=Paper=
{{Paper
|id=None
|storemode=property
|title=None
|pdfUrl=https://ceur-ws.org/Vol-1000/ICTERI-2013-CEUR-WS-Volume.pdf
|volume=Vol-1000
}}
==None==
https://ceur-ws.org/Vol-1000/ICTERI-2013-CEUR-WS-Volume.pdf
Vadim Ermolayev
Heinrich C. Mayr
Mykola Nikitchenko
Aleksander Spivakovskiy
Grygoriy Zholtkevych
Mikhail Zavileysky
Hennadiy Kravtsov
Vitaliy Kobets
Vladimir Peschanenko
(Eds.)
ICT in Education, Research
and Industrial Applications:
Integration, Harmonization
and Knowledge Transfer
Proceedings of the 9th International Conference,
ICTERI 2013
Kherson, Ukraine
June, 2013
Ermolayev, V., Mayr, H. C., Nikitchenko, M., Spivakovskiy, A., Zholtkevych, G.,
Zavileysky, M., Kravtsov, H., Kobets, V. and Peschanenko, V. (Eds.): ICT in Educa-
tion, Research and Industrial Applications: Integration, Harmonization and Knowled-
ge Transfer. Proc. 9th Int. Conf. ICTERI 2013, Kherson, Ukraine, June 19-22, 2013,
CEUR-WS.org, online
This volume constitutes the refereed proceedings of the 9th International Confer-
ence on ICT in Education, Research, and Industrial Applications, held in Kherson,
Ukraine, in June 2013.
The 49 papers were carefully reviewed and selected from 124 submissions. The
volume opens with the contributions of the invited speakers. Further, the part of the
volume containing the papers of the main ICTERI conference is structured in four
topical parts: ICT infrastructures, Integration and Interoperability; Machine Intelli-
gence, Knowledge Engineering and Management for ICT; Model-based software
system development; and Methodological and Didactical Aspects of Teaching ICT
and Using ICT in Education. This part of the volume is concluded by the two papers
describing the tutorials presented at the conference. The final part of the volume com-
prises the selected contributions of the three workshops co-located with ICTERI
2013, namely: the 1st International Workshop on Methods and Resources of Distance
Learning (MRDL 2013); the 2nd International Workshop on Information Technolo-
gies in Economic Research (ITER 2013); and the 2nd International Workshop on
Algebraic, Logical, and Algorithmic Methods of System Modeling, Specification and
Verification (SMSV 2013).
Copyright © 2013 for the individual papers by the papers’ authors.
Copying permitted only for private and academic purposes. This vol-
ume is published and copyrighted by its editors.
Preface
It is our pleasure to present you the proceedings of ICTERI 2013, the ninth edition of
the International Conference on Information and Communication Technologies in
Education, Research, and Industrial Applications: Integration, Harmonization, and
Knowledge Transfer, held at Kherson, Ukraine on June 19-22, 2013. ICTERI is con-
cerned with interrelated topics from information and communication technology
(ICT) infrastructures to teaching these technologies or using those in education or
industry. Those aspects of ICT research, development, technology transfer, and use in
real world cases are vibrant for both the academic and industrial communities.
The conference scope was outlined as a constellation of the following themes:
ICT infrastructures, integration and interoperability
Machine Intelligence, knowledge engineering, and knowledge management
for ICT
Cooperation between academia and industry in ICT
Model-based software system development
Methodological and didactical aspects of teaching ICT and using ICT in education
A visit to Google Analytics proves the broad and continuously increasing interest
in the ICTERI themes. Indeed, between November 15, 2012 and May 15, 2013 we
have received circa 4 400 visits to the conference web site, http://icteri.org/, from 110
countries (568 cities). These numbers are 1.5 – 2 times higher than those observed in
the similar period for ICTERI 2012.
ICTERI 2013 continues the tradition of hosting co-located focused events under its
umbrella. In the complement to the main conference this year, the program offered the
three co-located workshops, two tutorials, and IT talks panel. The main conference
program has been composed of the top-rated submissions evenly covering all the
themes of ICTERI scope.
The workshops formed the corolla around the main ICTERI conference by focus-
ing on particular sub-fields relevant to the conference theme. In particular:
The 1st International Workshop on Methods and Resources of Distance Learning
(MRDL 2013) dealt mainly with the methodological and didactical aspects of
teaching ICT and using ICT in education
The scope of the 2nd International Workshop on Information Technologies in Eco-
nomic Research (ITER 2013) was more within the topic of cooperation between
academia and industry
II
2nd International Workshop on Algebraic, Logical, and Algorithmic Methods of
System Modeling, Specification and Verification (SMSV 2013) focused on model-
based software system development
The IT Talks panel was the venue for the invited industrial speakers who wish to
present their cutting edge ICT achievements.
This year we were also accepted the two focused short tutorials to the program: on
ontology alignment and the industrial applications of this technology; and on the time
model and Clock Constraint Specification Language for the UML profile used in
modeling and analysis of real-time and embedded systems.
Overall ICTERI attracted a substantial number of submissions – a total of 124
comprising the main conference and workshops. Out of the 60 paper submissions to
the main conference we have accepted 22 high quality and most interesting papers to
be presented at the conference and published in our proceedings. The acceptance rate
was therefore 36.7 percent. Our three workshops received overall 64 submissions,
from which 27 were accepted by their organizers and included in the second part of
this volume. Those selected publications are preceded by the contributions of our
invited speakers. The talk by our keynote speaker Wolf-Ekkehard Matzke expressed
his industrial views on the knowledge-based bio-economy and the “Green Triple-
Helix” of biotechnology, synthetic biology, and ICT. The invited talk by Gary L. Pratt
was focused on a movement of higher education institutions to forming consortiums
for creating a position of strength facing contemporary economic challenges. The
invited talk by Alexander A. Letichevsky presented a general theory of interaction
and cognitive architectures based on this theory.
The conference would not have been possible without the support of many people.
First of all we would like to thank all the authors who submitted papers to ICTERI
2013 and thus demonstrated their interest in the research problems within our scope.
We are also very grateful to the members of our Program Committee for providing
timely and thorough reviews and also for been cooperative in doing additional review
work. We would like also to thank the local organizers of the conference whose devo-
tion and efficiency made this instance of ICTERI a very comfortable and effective
scientific forum. Finally a special acknowledgement is given to the support by our
editorial assistant Olga Tatarintseva who invested a considerable effort in checking
and proofing the final versions of our papers.
June, 2013 Vadim Ermolayev
Heinrich C. Mayr
Mykola Nikitchenko
Aleksander Spivakovskiy
Grygoriy Zholtkevych
Mikhail Zavileysky
Hennadiy Kravtsov
Vitaliy Kobets
Vladimir Peschanenko
III
Organization
Organizers
Ministry of Education and Science of Ukraine
Kherson State University, Ukraine
Alpen-Adria-Universität Klagenfurt, Austria
Zaporizhzhya National University, Ukraine
Institute of Information Technology and Teaching Resources, Ukraine
V. N. Karazin Kharkiv National University, Ukraine
Taras Shevchenko National University of Kyiv, Ukraine
DataArt Solutions Inc.
General Chair
Aleksander Spivakovsky, Kherson State University, Ukraine
Steering Committee
Vadim Ermolayev, Zaporizhzhya National University, Ukraine
Heinrich C. Mayr, Alpen-Adria-Universät Klagenfurt, Austria
Natalia Morse, National University of Life and Environmental Sciences, Ukraine
Mykola Nikitchenko, Taras Shevchenko National University of Kyiv, Ukraine
Aleksander Spivakovsky, Kherson State University, Ukraine
Mikhail Zavileysky, DataArt, Russian Federation
Grygoriy Zholtkevych, V.N.Karazin Kharkiv National University, Ukraine
Program Co-chairs
Vadim Ermolayev, Zaporizhzhya National University, Ukraine
Heinrich C. Mayr, Alpen-Adria-Universität Klagenfurt, Klagenfurt, Austria
IV
Workshops Chair
Mykola Nikitchenko, Taras Shevchenko National University of Kyiv, Ukraine
Tutorials Chair
Grygoriy Zholtkevych, V.N.Karazin Kharkiv National University, Ukraine
IT Talks Co-chairs
Aleksander Spivakovsky, Kherson State University, Ukraine
Mikhail Zavileysky, DataArt, Russian Federation
Program Committee
Rajendra Akerkar, Western Norway Research Institute, Norway
Eugene Alferov, Kherson State University, Ukraine
Costin Badica, University of Craiova, Romania
Tobias Buerger, PAYBACK, Germany
Andrey Bulat, Kherson State University, Ukraine
David Camacho, Universidad Autonoma de Madrid, Spain
Michael Cochez, University of Jyväskylä, Finland
Maxim Davidovsky, Zaporizhzhya National University, Ukraine
Anatoliy Doroshenko, National Technical University of Ukraine "Kyiv Polytechnic Institute”, Ukraine
Vadim Ermolayev, Zaporizhzhya National University, Ukraine
David Esteban, Techforce, Spain
Lyudmila Gavrilova, Slovyansk State Pedagogical University, Ukraine
Vladimir Gorodetsky, St. Petersburg Institute for Informatics and Automation of The Russian
Academy of Science, Russian Federation
Marko Grobelnik, Jozef Stefan Institute, Slovenia
Brian Hainey, Glasgow Caledonian University, United Kingdom
Sungkook Han, Wonkwang University, South Korea
Mitja Jermol, Jozef Stefan Institute, Slovenia
Jason Jung, Yeungnam University, South Korea
Natalya Keberle, Zaporizhzhya National University, Ukraine
Nick Kings, Connected Shopping, United Kingdom
Christian Kop, Alpen-Adria-Universität Klagenfurt, Austria
Hennadiy Kravtsov, Kherson State University, Ukraine
Nataliya Kushnir, Kherson State University, Ukraine
Frédéric Mallet, Université de Nice-Sophia Antipolis, France
Mihhail Matskin, Royal Institute of Technology, Sweden
Heinrich C. Mayr, Alpen-Adria-Universität Klagenfurt, Austria
Mykola Nikitchenko, Taras Shevchenko National University of Kyiv, Ukraine
Andriy Nikolov, Knowledge Media Institute, The Open University, United Kingdom
Inna Novalija, Jozef Stefan Institute, Slovenia
Tomás Pariente Lobo, ATOS Origin, Spain
Vladimir Peschanenko, Kherson State University, Ukraine
Carlos Ruiz, playence, Spain
Abdel-Badeeh Salem, Ain Shams University, Cairo, Egypt
Wolfgang Schreiner, Research Institute for Symbolic Computation (RISC),
Johannes Kepler University, Austria
V
Vladimir A. Shekhovtsov, Alpen-Adria-Universität Klagenfurt, Austria
Mikhail Simonov, Istituto Superiore Mario Boella, Italy
Marcus Spies, Ludwig-Maximilians-Universität München, Germany
Aleksander Spivakovsky, Kherson State University, Ukraine
Martin Strecker, IRIT, Universite Paul Sabatier, France
Olga Tatarintseva, Zaporizhzhya National University, Ukraine
Vagan Terziyan, University of Jyväskylä, Finland
Nikolay Tkachuk, National Technical University "Kharkiv Polytechnic Institute”, Ukraine
Leo Van Moergestel, Utrecht University of Applied Sciences, Netherlands
Maryna Vladymyrova, V. N. Karazin Kharkov National University, Ukraine
Paul Warren, Knowledge Media Institute, the Open University, United Kingdom
Iryna Zaretska, V. N. Karazin Kharkov National University, Ukraine
Mikhail Zavileysky, DataArt Solutions Inc., Russian Federation
Grygoriy Zholtkevych, V. N. Karazin Kharkov National University, Ukraine
Additional Reviewers
Fahdi Al Machot, Alpen-Adria-Universität Klagenfurt, Austria
Antonio Gonzalez-Pardo, Universidad Autonoma de Madrid, Spain
Alexey Vekschin, National Technical University "Kharkiv Polytechnic Institute”, Ukraine
VI
Sponsors
Organizations and Companies
DataArt (http://dataart.com/) develops custom applications,
helping clients optimize time-to-market and save costs
Kherson State University (http://www.ksu.ks.ua/) is a mul-
tidisciplinary scientific, educational, and cultural center in the
south of Ukraine
Zaporizhzhya National University (http://www.znu.edu.ua/)
is a renowned educational and research center of Ukraine that
offers a classically balanced diversity of high quality aca-
demic curricula and many opportunities to build your scien-
tific carrier
Individuals
Aleksandr Spivakovsky is the chair of the Department of
Informatics and the first vice-rector of Kherson State Univer-
sity
Dmitriy Shchedrolosev is the head of DataArt’s R&D Center
at Kherson
VII
Table of Contents
Preface ..........................................................................................................................I
Organization............................................................................................................. III
Sponsors.....................................................................................................................VI
Invited Contributions ................................................................................................. 1
The Knowledge-Based Bio-Economy and the “Green Triple-Helix” of
Biotechnology, Synthetic Biology and ICT .............................................................. 2
A Movement of Higher Education Institutions to Consortiums of Institutions
Banding Together to Create a Position of Strength.................................................... 3
General Theory of Interaction and Cognitive Architectures ...................................... 4
Part 1. Main ICTERI Conference ........................................................................... 16
1.1 ICT Infrastructures, Integration and Interoperability ......................................... 17
Wireframe Model for Simulating Quantum Information Processing Systems......... 18
Modeling, Algorithms and Implementation of the Microcontroller Control
System for the Ion Beam Forming Process for Nanostructures Etching.................. 30
Using Algebra-Algorithmic and Term Rewriting Tools for Developing
Efficient Parallel Programs ...................................................................................... 38
1.2 Machine Intelligence, Knowledge Engineering and Management for ICT........ 47
An Intelligent Approach to Increase Efficiency of IT-Service Management
Systems: University Case-Study.............................................................................. 48
Refining an Ontology by Learning Stakeholder Votes from their Texts ................. 64
Answering Conjunctive Queries over a Temporally-Ordered Finite
Sequence of ABoxes sharing one TBox................................................................... 79
An Adaptive Forecasting of Nonlinear Nonstationary Time Series
under Short Learning Samples................................................................................. 91
Application of an Instance Migration Solution to Industrial Ontologies ................ 99
Extracting Knowledge Tokens from Text Streams ................................................ 108
1.3 Model-Based Software System Development.................................................. 117
VIII
Use of Neural Networks for Monitoring Beam Spectrum of Industrial
Electron Accelerators............................................................................................. 118
Lazy Parallel Synchronous Composition of Infinite Transition Systems .............. 130
Selecting Mathematical Software for Dependability Assessment
of Computer Systems Described by Stiff Markov Chains ..................................... 146
Asymptotical Information Bound of Consecutive Qubit Binary Testing............... 163
A Data Transfer Model of Computer-Aided Vehicle Traffic
Coordination System for the Rail Transport in Ukraine ........................................ 178
Quantitative Estimation of Competency as a Fuzzy Set ........................................ 187
1.4 Methodological and Didactical Aspects of Teaching ICT
and Using ICT in Education............................................................................. 194
New Approaches of Teaching ICT to Meet Educational Needs
of Net Students Generation.................................................................................... 195
Pedagogical Diagnostics with Use of Computer Technologies ............................. 209
The Use of Distributed Version Control Systems in Advanced
Programming Courses............................................................................................ 221
Comparative Analysis of Learning in Three-Subjective Didactic Model .............. 236
Conception of Programs Factory for Representing and E-Learning
Disciplines of Software Engineering ..................................................................... 252
Public Information Environment of a Modern University ..................................... 264
Designing Massive Open Online Courses.............................................................. 273
The Role of Informatization in the Change of Higher School Tasks:
the Impact on the Professional Teacher Competences........................................... 281
1.5 ICTERI Tutorials ............................................................................................. 288
UML Profile for MARTE: Time Model and CCSL............................................... 289
Ontology Alignment and Applications in 90 Minutes ........................................... 295
Part 2. ICTERI Workshops ................................................................................... 307
2.1 2nd International Workshop on Information Technologies
in Economic Research (ITER 2013)................................................................ 308
Foreword................................................................................................................ 309
Binary Quasi Equidistant and Reflected Codes in Mixed Numeration Systems.... 311
Mechanism Design for Foreign Producers of Unique Homogeneity Product........ 329
Features of National Welfare Innovative Potential Parametric Indication
Information-Analytical Tools System in the Globalization Trends’ Context ........ 339
Matrix Analogues of the Diffie-Hellman Protocol ................................................ 352
Are Securities Secure: Study of the Influence of the International
Debt Securities on the Economic Growth.............................................................. 360
How to Make High-tech Industry Highly Developed? Effective Model
of National R&D Investment Policy...................................................................... 366
IX
Econometric Analysis on the Site “Lesson Pulse”................................................. 374
Decision Supporting Procedure for Strategic Planning: DEA Implementation
for Regional Economy Efficiency Estimation ....................................................... 385
Applying of Fuzzy Logic Modeling for the Assessment
of ERP Projects Efficiency .................................................................................... 393
Mathematical Model of Banking Firm as Tool for Analysis,
Management and Learning .................................................................................... 401
2.2 1st International Workshop on Methods and Resources
of Distance Learning (MRDL 2013) ................................................................ 409
Foreword................................................................................................................ 410
What Should be E-Learning Course for Smart Education ..................................... 411
TIO – a Software Toolset for Mobile Learning in MINT Disciplines .................. 424
Holistic Approach to Training of ICT Skilled Educational Personnel................... 436
2.3 2nd International Workshop on Algebraic, Logical, and Algorithmic Methods
of System Modeling, Specification and Verification (SMSV 2013) ................ 446
Foreword................................................................................................................ 447
An Abstract Block Formalism for Engineering Systems ....................................... 448
Multilevel Environments in Insertion Modeling System ....................................... 464
Clocks Model for Specification and Analysis of Timing
in Real-Time Embedded Systems .......................................................................... 475
Specializations and Symbolic Modeling................................................................ 490
On a Dynamic Logic for Graph Rewriting ............................................................ 506
Logical Foundations for Reasoning about Transformations
of Knowledge Bases .............................................................................................. 521
Program Algebras with Monotone Floyd-Hoare Composition .............................. 533
A Formal Model of Resource Sharing Conicts in Multithreaded Java .................. 550
Implementation of Propagation-Based Constraint Solver in IMS.......................... 565
UniTESK: Component Model Based Testing........................................................ 573
Protoautomata as Models of Systems with Data Accumulation ............................ 582
Models of Class Specification Intersection of Object-Oriented Programming...... 590
Author Index ........................................................................................................... 595
X
Invited Contributions
The Knowledge-Based Bio-Economy and the “Green
Triple-Helix” of Biotechnology, Synthetic Biology
and ICT
Wolf-Ekkehard Matzke
MINRES Technologises GmbH, Neubiberg, Germany
wolf@minres.com
Abstract. Over the last decades economies around the globe have transformed
into a knowledge-based economy (KBE). Information and Communication
Technology (ICT) has served as the principal enabling technology for this trans-
formation. Now biology becomes another major pillar – producing a knowl-
edge-based bio-economy (KBBE). The challenges faced by biotechnology push
the requirements for ICT in many ways to the extreme and far beyond its basic
utility function. In particular, it is valid for synthetic biology which aims to
break ground on the rational design and construction of artificial biological sys-
tems with ICT as its backbone for bio-design automation (BDA). This could be
best illustrated using a metaphor of a “green triple-helix”, where “green” stands
for environmental consciousness and “triple-helix” visualizes the inter-
dependency of biotechnology, synthetic biology, and ICT as the helical strands.
The talk will explore this inter-dependency in dynamics. High level ICT re-
quirements will be identified and discussed along the dimensions of education,
research and industry with the emphasis on synthetic biology and BDA. The
guidelines for the architecture and implementation of an open BDA platform
will be presented so that interested ICT researchers and practitioners will better
understand the biology-specific ICT challenges of the KBBE.
Keywords. Knowledge-Based Bio-Economy, Biotechnology, Synthetic Biol-
ogy, Bio-Design Automation
Key terms. ICTInfrastructure, Industry, Management, Research
A Movement of Higher Education Institutions
to Consortiums of Institutions Banding Together
to Create a Position of Strength
Gary L. Pratt
Eastern Washington University, 202 Huston Hall, Cheney, Washington 99004, USA
gpratt@ewu.edu
Abstract. Colleges and universities compete for students, faculty, and business,
industry, and research partnerships with quality programs, strong faculty, re-
search opportunities, affordable cost, and high student success factors. Yet, at
the infrastructure level, most of these institutions provide many similar informa-
tion technology services and support. On top of this, many of these institutions
struggle to provide this quality infrastructure because of a variety of factors, in-
cluding: shrinking budgets, minimal strategic planning, and a lack of institu-
tional vision of information technology as a strategic asset. This presentation
will showcase best practice examples of how higher education institutions can
band together, to create strong consortium relationships that can help all part-
ners in this relationship move forward as a strong force. Examples will include
actual successes experience by the Kentucky Council on Postsecondary Educa-
tions Distance Learning Advisory Committee (DLAC), the Washington Legis-
lative Technology Transformation Taskforce (TTT), and the Washington
Higher Education Technology Consortium (WHETC).These successes range
from statewide strategic planning efforts, to significant consortial purchasing
contracts, to collaborative technology systems, services, and training opportuni-
ties. This presentation will show that institutions can be stronger working to-
gether than working individually.
Keywords. University consortium, best practice, competition, infrastructure, in-
formation technology, strategic asset, strategic planning, collaborative technol-
ogy system
Key terms. Academia, Information Technology, Infrastructure, Cooperation,
Management
General Theory of Interaction and Cognitive
Architectures
Alexander Letichevsky
Glushkov Institute of Cybernetics, Academy of Sciences of Ukraine
40 Glushkova ave., 03187, Kyiv, Ukraine
let@cyfra.net
Abstract. The challenge of creating a real-life computational equiva-
lent of the human mind is now attracting the attention of many scientific
groups from different areas of cybernetics and Artificial Intelligence such
as computational neuroscience, cognitive science, biologically inspired
cognitive architectures etc. The paper presents a new cognitive archi-
tecture based on insertion modeling, one of the paradigms of a general
theory of interaction, and a basis for multiagent system development.
Insertion cognitive architecture is represented as a multilevel insertion
machine which realizes itself as a high level insertion environment. It has
a center to evaluate the success of its behavior which is a special type
agent that can observe the interaction of a system with external environ-
ment. The main goal of a system is achieving maximum success repeated.
As an agent this machine is inserted into its external environment and
has the means to interact with it. The internal environment of intelligent
cognitive agent creates and develops its own model and the model of ex-
ternal environment. If the external environment contains other agents,
they can be modeled by internal environment which creates correspond-
ing machines and interprets those machines using corresponding drivers,
comparing the behaviors of models and external agents. Insertion ar-
chitecture is now under development on the base of Insertion modeling
system, developed in Glushkov Institute of Cybernetics.
Keywords. AgentBasedSystem, DistributedArtificialIntelligence, Rea-
soning, FormalMethod, Simulation
Key terms. AgentBasedSystem, DistributedArtificialIntelligence, Rea-
soning, FormalMethod, Simulation
1 Introduction
General theory of interaction is a theory of information interaction in complex
distributed multi-agent systems. It has a long history. Contemporary part of
this history can be considered as starting from neuro networks of McCulloch-
Pitts [23]. The model of neuro nets caused the appearance of abstract automata
theory, a theory which helps study the behavior and interaction of evolving
systems independently of their structure. The Kleene-Glushkov algebra [13, 7]
General Theory of Interaction and Cognitive Architectures 5
is the main tool for the description of the behaviors of finite state systems.
Automata theory originally concentrated on the study of analyses and synthesis
problems, generalization of finite state automata and complexity. Interaction in
explicit form appeared only in 70s as a general theory of interacting information
processes. It includes the CCS (Calculus of Communicated Processes) [24, 25]
and the π-calculus of R. Milner [26], CSP (Communicated Sequential Processes)
of T. Hoare [10], ACP (Algebra of Communicated Processes) [3] and many other
various branches of these basic theories. Now all these calculi and algebras are
the basis for modern research in this area. Fairly complete survey of the classical
process theory is presented in the Handbook of Process Algebras [4], published
in 2001.
Insertion modeling is a trend that is developing over the last decade as an
approach to a general theory of interaction of agents and environments in com-
plex distributed multi-agent systems. The first works in this direction have been
published in the middle of 90s [6, 15, 16]. In these studies, a model of interac-
tion between agents and environments based on the notion of insertion function
and the algebra of behaviors (similar to some kind of process algebra) has been
proposed. The paradigm shift from computing to interaction was extensively dis-
cussed in computer science that time, and our work was in some sense a response
to this trend. But the real roots of the insertion model should be sought even
earlier, in a model of interacting of control and operational automata, proposed
by V. Glushkov back in the 60s [8, 9] to describe the structure of computers.
In the 70s the algebraic abstraction of this model were studied in the theory of
discrete processors and provided a number of important results on the problem
of equivalence of programs, their equivalent transformations and optimization.
Macroconveyor models of parallel computing, which were investigated in 80s
years [11], even more close to the model of interaction of agents and environ-
ments. In these models, the processes corresponding to the parallel processors
can be considered as agents that interact in an environment of distributed data
structures.
In recent years, insertion modeling has been applied to the development
of systems for the verification of requirements and specifications of distributed
interacting systems [2, 12, 19–21]. The system VRS, developed in order from Mo-
torola, has been successfully applied to verify the requirements and specifications
in the field of telecommunication systems, embedded systems, and real-time sys-
tems. A new insertion modeling system IMS [17], which is under development
in the Glushkov Institute of Cybernetics of the National Academy of Sciences
of Ukraine, is intended to extend the area of insertion modeling applications.
We found many common features of the tools used in software development area
based on formal methods and techniques used in biologically inspired cognitive
architectures. This gives us hope to introduce some new ideas to the development
of this subject domain.
This paper presents the main principals of insertion modeling and the con-
ception of cognitive architecture based on insertion modeling. To understand the
formal part of the paper reader must be familiar with the concepts of labeled
6 A. Letichevsky
transition system, bisimilarity and basic notions of general process theory. The
mathematical foundation of insertion modeling is presented in [18].
2 The Basic Principals
Insertion modeling deals with the construction of models and study the inter-
action of agents and environments in complex distributed multi-agent systems.
Informally, the basic principles of the paradigm of insertion modeling can be
formulated as follows.
1. The world is a hierarchy of environments and agents inserted into them.
2. Environments and agents are entities evolving in time.
3. Insertion of agent into environment changes the behavior of environment and
produces new environment which is in general ready for the insertion of new
agents.
4. Environments as agents can be inserted into higher level environment.
5. New agents can be inserted from external environment as well as from in-
ternal agents (environments).
6. Agents and environments can model another agents and environments on
the different levels of abstraction.
All these principles can be formalized in terms of transition systems, behavior
algebras, and insertion functions. This formalization can be used as high level
abstractions of biological entities needed for computer modeling of human mind.
The first and the second principals are commonly used in information mod-
elling of different kinds of systems, for example as in object oriented or agent
programming. They are also resembling to M. Minsky’s approach of the society
of mind [27].
The third principal is clear intuitively, but has a special refinement in inser-
tion modelling. We treat agents as transition systems with states considered up
to bisimilarity (or up to behavior, which is the same). The type of an agent is
the set of actions it can perform. The term action we use as a synonim of label
for transitions, and it can denote signals or messages to send, events in which an
agent can participate etc. This is the most general notion of agent which must
be distinguished from more special notions of autonomous or intellectual agents
in AI.
Transition system consists of states and transitions that connect states. Tran-
sitions are labeled by actions (signals, events, instructions, statements etc.).
Transition systems are evolving in time changing their states, and actions are
observable symbolic structures used for communication. We use the well-known
a
notation s −→ s0 to express the fact that transition system can evolve from the
0
state s to s performing action a. Usually transition systems are nondeterministic
and there can be several transitions coming from the same state even labeled by
the same action. If we abstract from the structure of states and concentrate only
on (branching) sequences of observable actions we obtain the state equivalence
General Theory of Interaction and Cognitive Architectures 7
called bisimilarity (originated from [28] and [24], exact definition can be found
in [18]). Bisimilar states generate the same behavior of transition systems.
Environment by definition is an agent that possesses the insertion function.
Given the state of environment s and the state of agent u, insertion function
computes the new state of environment which is denoted as s[u]. Note that we
consider states up to bisimilarity and if we have some representation of behaviors,
the behaviors of environment and agent can be used as states. The state s[u]
is a state of environment and we can use insertion function to insert a new
agent v into environment s[u] : (s[u])[v] = s[u, v]. Repeating this construction
we can obtain the state of environment s[u1 , u2 , . . .] with several agents inserted
into it. Insertion function can be considered as an operator over the states of
environment, and if the states are identified with behaviors, then the insertion
of a new agent changes the behavior of environment.
Environment is an agent with insertion function, so if we forget the inser-
tion function, then environment can be inserted as agent into a higher level
environment and we can obtain hierarchical structure like
s[s1 [u11 , u12 , . . .]E1 , s2 [u21 , u22 , . . .]E2 , . . .]E
Here notation s[u1 , u2 , . . .]E explicitly shows the environment E to which the
state s belongs (environment indexes can be omitted if they are known from the
context). This refines the fourth principle.
Environment is an agent which can be inserted into external environment
and having agents inserted into this environment. The evolution of agents can
a a
be defined by the rules for transitions. The rules s[u] −→ s[u, v] and s[t[u, v]] −→
s[t[u], v] can be used for the illustration of the 5-th principal.
We consider the creating and manipulation of the models of external and
internal environments as the main property of cognitive processes of intellectual
agent. Formalization of this property in terms of insertion modeling supports
the 6-th principal.
Cognitive architecture will be constructed as a multilevel insertion environ-
ment. Below we shall define the main kinds of blocks used for construction of
cognitive architecture. They are local description unites and insertion machines.
3 Multilevel Environments
To represent behaviors of transition systems we use behavior algebras (a kind of
process algebra). Behavior algebra is defined by the set of actions and the set
of behaviors (processes). It has two operations and termination constants. Op-
erations are prefixing a.u (a - action, u - behavior) and nondeterministic choice
u + v (u and v - behaviors). Termination constants are successful termination
∆, deadlock 0, and undefined behavior ⊥. It has also approximation relation ⊆,
which is a partial order with minimal element ⊥, and is used for constructing
a complete algebra with fixed point theorem. To define infinite behaviors we
use equations in behavior algebra. These equations have the form of recursive
definitions ui = Fi (u1 , u2 , . . .), i = 1, 2, . . . and define left hand side functions as
8 A. Letichevsky
the components of a minimal fixed point. Left hand sides of these definitions can
depend on parameters ui (x) = Fi (u, x) of different types. In complete behavior
algebra each behavior has a representation (normal form)
X
u= ai .ui + εi
i∈I
which is defined uniquely (up to commutativity and associativity of nondeter-
ministic choice), if all ai .ui are different (εu is a termination constant).
The type of environment is defined by two action sets: the set of environment
actions and the set of agent actions. The last defines the type of agents which
can be inserted into this environment: if the set of agent actions is included in
the set of agent actions of environment then this agent can be inserted into this
environment. This relation is called compatibility relation between agents and
environments (agent is compatible with environment if it can be inserted into
this environment). Multilevel environment is a family of environments with dis-
tinguished the most external environment. The compatibility relation on the set
of environments defines a directed graph and we demand for multilevel environ-
ment that the outermost environment would be reachable from any environment
of the family in this graph.
To define the insertion function for some environment it is sufficient to define
transition relation for all states of environment including states with inserted
agents. The common approach is to define behavior by means of rules. The
following is an example of such rule:
b a
− s0 , u −
s→ → u0
c P (a, b, c)
− s0 [u0 ]
s[u] →
This rule can be interpreted as follows. Agent in the state u can make a transition
a
u−→ u0 . Environment allows this transition if the predicate P (a, b, c) is true. This
rule defines behavior property of environment in some local neighborhood of the
state s[u]. So such a rule belongs to the class of local description units discussed
in the next section.
At a given moment of time an agent belongs (is inserted) to only one en-
vironment. But if the type of an agent is compatible with the type of another
environment it can move to this environment. Such a movements can be de-
scribed by the following types of rules:
moveup E
u −−−−−−−→ u0
P1 (E, F, u, moveup(E))
moveup(F →E)
E[F [u, v], w] −−−−−−−−−−→ E[F [v], u0 , w]
moving from internal to external environment;
movedn F
u −−−−−−−→ u0
P2 (E, F, u, movedn(F ))
movedn(E→F )
E[F [v], u, w] −−−−−−−−−−→ E[F [u0 , v], w]
General Theory of Interaction and Cognitive Architectures 9
moving from external environment to internal one;
moveto G
u −−−−−−→ u0
P3 (E, F, u, moveto(F ))
moveto(F →G)
E[F [u, v], G[w]] −−−−−−−−−→ E[F [v], G[u0 , w]]
moving to another environment on the same level. In all cases permitting con-
ditions must include the compatibility conditions for corresponding agents and
environments. The rules above define the property of a system called mobility
and underlies the calculus of mobile ambients of Luca Cardelli [5].
4 Local Description Units over Attribute Environments
A special type of environments is considered in cognitive architecture to have
a sufficiently rich language for the description of environment states properties.
These environments are called attribute environments. There are two kinds of
attribute environments – concrete and symbolic.
The state of concrete attribute environment is the valuation of attributes -
symbols that change their values while changing the state in time. Each attribute
has type (numeric, symbolic, enumerated, agent and behavior types, functional
types etc.). Some of functional and predicate symbols are interpreted symbols.
Now logic formulas can be used for the description of properties of agent or
environment states. We use the first order logic formulas as the basis that can
be extended by fuzzy logic, temporal logic etc.
The general form of local description unit is the following:
∀x(α(x, r) →< P (x, r) > β(x, r)),
where x is a list of typed parameters, r is a list of attributes, α(x, r) and β(x, r)
are logic formulas, < P (x, r) > is a process - finite behavior of an environment.
Local descriptions can be considered as formulas of dynamic logic, or Hoare
triples, or productions - the most popular units of procedural memory in AI.
In any case they describe local dynamic properties of environment behavior:
for all possible values of parameters, if precondition is true then a process of a
local description unit can start and after successful termination of this process
a postcondition must be true.
The states of symbolic environment are formulas of basic logic language of
environment. Such formulas are abstractions of classes of concrete states. Each
symbolic state covers the set of concrete states and the traces generated by local
description units cover the sets of concrete traces.
Local description units are the main units of knowledge representation in cog-
nitive architecture. A set of local description units can be used for the definition
of transitions of environment. In this case they can be considered as procedural
knowledge units. Logic knowledge can be represented as environment with the
states representing the current knowledge, and the local description units cor-
responding to the rules of inference in corresponding calculus. Local description
units can be applied in forward and backward modes. Forward mode can be used
for the generating of new knowledge, backward mode – for answering queries.
10 A. Letichevsky
5 Insertion Machines
Another construction blocks for cognitive architecture are insertion machines
intended for implementation of insertion environments. The input of insertion
machine is the description of a multilevel environment (a model of an environ-
ment) and its initial state, an output depends on the goal that is put to machine.
Multilevel environments are represented in cognitive architecture by means
of environment descriptions for different levels and a set of local description
units for insertion functions. Environment description contains the signature of
environment that includes types of attributes, types of inserted agents, and also
the description of sets of environment and agent actions. Local description units
used for the definition of insertion function are organized as a knowledge base
with special data structures providing efficient access to the needed descriptions
and history of their use.
To implement multilevel environment different kinds of insertion machines
are used. But all of them have the general architecture represented on the Fig.1.
Three main components of insertion machine are model driver (MD), behavior
Fig. 1. Architecture of Insertion Machine
unfolder (Unf), and interactor (Intr). Model driver is a component which con-
trols the machine traversal along the behavior tree of a model. The state of a
model is represented as a text in the input language of insertion machine and is
General Theory of Interaction and Cognitive Architectures 11
considered as an algebraic expression. The input language includes the recursive
definitions of agent behaviors, the notation for insertion function, and possibly
some compositions for environment states. Before computing insertion function
the state of a system must be represented in the form s[u1 , u2 , . . .]. This func-
tionality is performed by agent behavior unfolder. To make the movement, the
state of environment must be reduced to the normal form
X
ai .ui + ε
i∈I
where ai are actions, ui are environment states, ε is a termination constant. This
functionality is performed by the module environment interactor. It computes
the insertion function calling recursively if it is necessary the agent behavior
unfolder.
Two kinds of insertion machines are distinguished: real time or interactive
and analytical insertion machines. The first ones are functioning in the real or
virtual environment, interacting with it in the real or virtual time. Analytical
machines intended for model analysis, investigation of its properties, solving
problems etc. The drivers for two kinds of machines correspondingly are also di-
vided into interactive and analytical drivers. Interactive driver after normalizing
the state of environment must select exactly one alternative and perform the
action specified as a prefix of this alternative. Insertion machine with interactive
driver operates as an agent inserted into external environment with insertion
function defining the laws of functioning of this environment. External environ-
ment, for example, can change a behavior prefix of insertion machine according
to their insertion function. Cognitive interactive driver has criteria of successful
functioning in external environment, it accumulates the information about its
past in episodic memory, develops the models of external environment, uses some
learning algorithms to improve the strategy of selecting actions and increase the
level of successful functioning. In addition it should have specialized tools for ex-
change the signals with external environment (for example, perception of visual
or acoustical information, space movement etc.).
Analytical insertion machine as opposed to interactive one can consider dif-
ferent variants of making decisions about performed actions, returning to choice
points (as in logic programming) and consider different paths in the behavior
tree of a model. The model of a system can include the model of external en-
vironment of this system, and the driver performance depends on the goals of
insertion machine. In the general case analytical machine solves the problems by
search of states, having the corresponding properties (goal states) or states in
which given safety properties are violated. The external environment for inser-
tion machine can be represented by a user who interacts with insertion machine,
sets problems, and controls the activity of insertion machine. Analytical machine
enriched by logic and deductive tools are used for generating traces of symbolic
models of systems. The state of symbolic model is represented by means of prop-
erties of the values of attributes rather than their concrete values.
Insertion machine with separated external environment interface can be im-
plemented as a transition system with hidden structure that separates the kernel
12 A. Letichevsky
environment state and the states of inserted agents. Such implementation can
be more efficient and can be constructed using partial computations or other
specialization and optimization programming tools.
6 Cognitive Architecture
Like well-known cognitive architectures such as Soar [14], ACT-R [1] or many
other from the list of BICA society [29] insertion cognitive architecture ICAR
is an environment for construction of cognitive agents. The main blocks of this
architecture are local description units, agents, represented by their behaviors,
and insertion machines. Building blocks are collected in memory units that have
structures of knowledge bases or associative memories.
On the abstract level ICAR has the same architecture as cognitive agents that
can be created in it. From this point of view it can be considered as an intellectual
assistant for user who interacts with ICAR in the process of creating cognitive
agents. The general architecture of cognitive agent of ICAR is represented on
Fig.2.
Fig. 2. Insertion cognitive architecture
In general cognitive agent is constructed as a real time multilevel insertion
machine which realizes itself as a highest level internal environment. As an agent,
General Theory of Interaction and Cognitive Architectures 13
this machine is inserted in its external environment and has the means to interact
with it. This external environment includes a user and objects of external (real
or virtual) world to which agent has access.
One or several self-models can be inserted into the internal environment of
cognitive agent to be used when interacting with external environment or making
decisions and planning future activities. An agent has an estimation mechanism
to evaluate the success of its behavior. This mechanism is realized in the form
of a special agent that can observe the interaction of a system with external
environment and make estimation according to some criteria. These criteria can
be predefined initially and evolves in the future according to obtained experience.
The main goal of a system is achieving maximum success repeated.
The self-models of cognitive agent are created and developed together with
the models of external environment. If the external environment contains other
agents, they can be modeled by internal environment which creates correspond-
ing machines and interprets those machines using corresponding drivers, com-
paring the behaviors of models and external agents. All these models are evolving
and developing in the process of accumulating the experience in interaction with
the external world.
Some mechanisms that model emotional or psychological features (humor and
concentration, pleasure and anger, etc.) can be implemented at higher levels of
cognitive strucure. The mechanisms of decision making, planning and executing
plans are also at higher levels.
The main part of cognitive structure is the base of models describing the
history of cognitive agent functioning at different levels of abstraction. The in-
terface with external world provides language (symbolic) communication and
image processing. All interaction histories are processed in the working memory
of the self-level insertion machines and then transferred to the appropriate levels
of a model base.
The model base is always active. The analytical insertion machines which
control and manage the structure of model base are always busy with searching
solution of problems and performing tasks with ansatisfactory answers, or cre-
ating new models. All this activity models subconcious levels of cognition and
time-to-time interact with the higher levels of cognitive structure. Independent
levels of cognitive structure are working in parallel.
The hierarchy of environments of cognitive agent in some sense are similar
to six layers of neocortex. Moving from low levels to higher ones the levels of
abstraction are increased and used more and more abstract symbolic models.
How to create such models is a big challenge and we are working on it now.
Cognitive analytical insertion machines of ICAR are used by cognitive agents
to learn their models and their interaction with external environment to solve
problems better, accepts user helps as a teacher and teach user how to interact
better with ICAR. General learning mechanisms are the parts of model drivers
of different types.
14 A. Letichevsky
7 Conclusions
The description of cognitive architecture in the last section is a very tentative
reflection of our far goals. The nearer goals include the further development
of our system of proving pogram correctness [22], communication in natural
language, and living in virtual reality. As a zero approximation of ICAR the
insertion modeling system [17] is used.
References
1. Anderson, J.R., Lebiere, C.: The Atomic Components of Thought. Mahwah:
Lawrence Erlbaum Associates (1998)
2. Baranov, S.,Jervis, C., Kotlyarov, V., Letichevsky, A. and Weigert, T.: Leveraging
UML to deliver correct telecom applications in UML for Real: Design of Embed-
ded Real-Time Systems by L.Lavagno, G. Martin, and B. Selic (editors), 323-342,
Kluwer Academic Publishers (2003)
3. Bergstra, J. A. and Klop J. W.: Process algebra for synchronous communications.
Information and Control, 60 (1/3), 109–137 (1984)
4. Bergstra, J. A., Ponce, A. and Smolka, S. A.(eds.): Handbook of Process Algebra.
North-Holland (2001)
5. Cardelli, L. and Gordon, A. D.: Mobile ambients. In: Foundations of Software
Science and Computation Structures: First International Conference, FOSSACS
’98, Springer-Verlag (1998)
6. Gilbert, D. R. and Letichevsky, A. A.: A Universal Interpreter for Nondetermin-
istic Concurrent Programming Languages. In: Gabbrielli, M. (Ed.) Fifth Compu-
log Network Area Meeting on Language Design and Semantic Analysis Methods,
September (1996)
7. Glushkov, V.M.: On an Algorithm of Abstract Automata Synthesis. Ukrainian
Mthematical Journal, 12(2), 147–156 (1960).
8. Glushkov, V.M.: Automata Theory and Questions of Design Structure of Digital
Machines. Cybernetics 1, 3–11 (1965)
9. Glushkov, V.M. and Letichevsky, A. A.: Theory of Algorithms and Discrete proces-
sors. In: Tou, J. T. (Ed.) Advances in Information Systems Science, vol. 1, Plenum
Press, 1-58 (1969)
10. Hoare, C.A.R.: Communicating Sequential Processes. Prentice Hall (1985)
11. Kapitonova, J. and Letichevsky, A.: Mathematical Theory of Computational Sys-
tems Design. Moscow, Science (1988) (in Russian)
12. Kapitonova, J., Letichevsky, A., Volkov, V. and Weigert, T.: Validation of Em-
bedded Systems. In: R. Zurawski (Ed.) The Embedded Systems Handbook, CRC
Press, Miami (2005)
13. Kleene, S. C.: Representation of Events in Nerve Nets and Finite Automata. In:
Shannon, C. E., McCarthy, J. (eds.) Automata Studies, Princeton University Press,
pp. 3-42 (1956)
14. Laird, J. E., Newell, A., Rosenbloom, P. S.: SOAR: an Architecture for General
Intelligence. Artifitial intelligence, 33, 1–64 (1987)
15. Letichevsky, A. and Gilbert, D.: A general Theory of Action Languages. Cyber-
netics and System Analyses, 1 (1998)
General Theory of Interaction and Cognitive Architectures 15
16. Letichevsky, A. and Gilbert, D.: A Model for Interaction of Agents and Environ-
ments. In: Bert, D., Choppy, C. and Moses, P. (eds.) Recent Trends in Algebraic
Development Techniques. LNCS, vol 1827, Springer Verlag (1999)
17. Letichevsky, A., Letychevskyi, O.and Peschanenko, V.: Insertion Modeling System.
In: Proc. PSI 2011, LNCS, vol 7162, pp. 262–274, Springer Verlag (2011)
18. Letichevsky, A.: Algebra of Behavior Transformations and its Applications. In:
Kudryavtsev, V. B. and Rosenberg, I.G. (eds.) Structural Theory of Automata,
Semigroups, and Universal Algebra. NATO Science Series II. Mathematics, Physics
and Chemistry, vol 207, pp. 241–272, Springer Verlag (2005)
19. Letichevsky, A., Kapitonova, J., Letichevsky, A. Jr., Volkov, V., Baranov, S., Kotl-
yarov, V. and Weigert, T.: Basic Protocols, Message Sequence Charts, and the
Verification of Requirements Specifications. ISSRE 2004, WITUL (Workshop on
Integrated reliability with Telecommunications and UML Languages) , Rennes, 4
November (2005)
20. Letichevsky, A., Kapitonova, J., Letichevsky, A. Jr., Volkov, V., Baranov, S., Kotl-
yarov, V. and Weigert, T.: Basic Protocols, Message Sequence Charts, and the Ver-
ification of Requirements Specifications. Computer Networks, 47, 662–675 (2005)
21. Letichevsky, A., Kapitonova, J., Letichevsky, A. Jr., Volkov, V., Baranov, S., Kotl-
yarov, V. and Weigert, T.: System Specification with Basic Protocols. Cybernetics
and System Analyses, 4 (2005)
22. Letichevsky, A., Letichevsky, O., Morokhovets, M. and Peschanenko, V.: System
of Programs Proving. In: Velichko, V., Volosin, A. and Markov, K. (eds.) Problems
of Computer Intellectualization, Kyiv, V. M. Glushkov Institute of Cybernetics,
pp.133–140 (2012)
23. McCulloch, W.S. and Pitts, W.: A Logical Calculus of the Ideas Immanent in
Nervous Activity, Bull. of Math Biophy., 5, 115–133 (1943)
24. Milner, R.: A Calculus of Communicating Systems, LNCS, vol 92, Springer Verlag
(1980)
25. Milner, R.: Communication and Concurrency. Prentice Hall (1989)
26. Milner, R.: The Polyadic π-calculus: a Tutorial. Tech. Rep. ECS–LFCS–91–180,
Laboratory for Foundations of Computer Science, Department of Computer Sci-
ence, University of Edinburgh, UK (1991)
27. Minsky, M.: The Society of Mind. Touchstone Book (1988)
28. Park, D.: Concurrency and Automata on Infinite Sequences. LNCS, vol 104,
Springer-Verlag (1981)
29. Samsonovich, A. V.: Toward a Unified Catalog of Implemented Cognitive Archi-
tectures (Review). In: Samsonovich, A.V., Johansdottir, K.R., Chella, A. and Go-
ertzel, B. (eds.) Biologically Inspired Cognitive Architectures 2010: Proc. 1st An-
nual Meeting of BICA Society, Frontiers in Artifitial Intelligence and Applications,
vol 221, pp. 195–244 (2010)
Part 1. Main ICTERI Conference
1.1 ICT Infrastructures, Integration
and Interoperability
Modeling, Algorithms and Implementation of the
Microcontroller Control System for the Ion Beam
Forming Process for Nanostructures Etching
Aleksandr Ralo1, Andrii Derevianko1, Aleksandr Kropotov1, Sergiy Styervoyedov1
and Oleksiy Vozniy1
1
V.N. Karazin Kharkiv National University,
Svobody sq. 4, 61022, Kharkiv, Ukraine
ralo-a-n@mail.ru
Abstract. In this work a three-level control system for the vacuum-plasma sys-
tem with the ion source based on high-frequency induction discharge architec-
ture is described. The system is built on smart sensors and specially designed
for operation in high EMI programmable logic controllers (PLCs) in the middle
hierarchical level. The structure of PLC and the algorithms of the system are
presented. Results of comparative simulation of classical management system
and created one, as well as the results of the system applying to obtain beams of
positive and negative ions and beams neutralized particles are reported.
Keywords. Programmable logic controller (PLC), control system, plasma tech-
nology, empirical model
Key terms. InformationTechnology, CommmunicationTechnology, Software-
System, Integration, Process
1 Introduction
Ion beams are an effective means of the surface treatment. Their application can be
found in the industry of the integrated circuits manufacturing, as well as for research
purposes as a powerful tool of micro-and nanoscale structures synthesis. However,
during the materials treatment by ion beams due to effects of similar charges repul-
sion and the charge accumulation near the processed surface, defects are formed in
the form of islet unetched films and vice versa – side etches. There are unwanted and
unpredictable changes in the structure of the processed sample that can irreversibly
change the characteristics of the ware [1]. There are several methods of excluding or
compensating the charge accumulation. One of the most promising technique that can
increase the rate of applicable products during plasma-beam processing is alternately
etching by pulsed beams of ions of different signs and etching by high-energy beams
Modeling, Algorithms and Implementation of the Microcontroller Control System 31
of neutral atoms, where the particle charge is neutralized by the ion beam away from
the treated sample [2], [3], [4].
Short times and complex algorithms that should be taken into account during these
experiments do not allow the operator to process without the use of modern computer
automation systems that work using a process model. Therefore, the aim of this work
was to create the intellectual management and control system, that will meet the re-
quirements of operating conditions of the pulsed plasma-process plant for etching and
micro- and nanostructures formation. An urgent task today is to build a system capa-
ble of independently analysis of historical data, and using them to provide the process
control. Seeing the specifics of the pulsed plasma-ion process system must ensure the
fastest possible response to the effects of interference that is impossible with the use
of information-analytical systems (IAS) based on the classical scheme and used for
the slower process.
2 The Use of Programmable Logic Controllers in Control
Systems
Programmable logic controllers (PLCs) are widely used in such control systems as
industrial automation and automation of scientific research. They provide signifi-
cantly higher reliability than personal computers, and the same flexibility of working.
Difficulties with changes in the operation make popular today microcontroller sys-
tems as not sufficient automation tool.
In order to improve the efficiency of the system it was proposed to build the hard-
ware using three-tier architecture, as it shown in Fig. 1. The lower level consists of
intelligent sensors and control elements, which control the PLC work.
PLCs have a much greater speed and are responsible for getting data from sensors,
filtering, formation of data packets and for communicating with the control centre,
created on the basis of a personal computer.
Algorithmic work of such system can be represented as follows:
1. Information about physical parameters, obtained by sensors, after conversion into
digital codes enters with the corresponding interface to the programmable logic
controller.
2. PLC forms an information packet, transmits the received information to the manag-
ing node (in our case, the system arbiter).
3. The central node addresses for the data needed to the historical data storage, ana-
lyzes them and, if necessary, generate management command. Information re-
ceived from the PLC is also stored in the repository for further analysis.
4. This control command is received with the PLC, which transmits it to the appropri-
ate control element.
This design exceeds the performance of classical one, because: first, damaged or
incorrect data can be filtered by a PLC, and second, the PLC data is transferred using
an optimum package format.
Modern PLC consists of three main parts:
PLC processor module
32 A. Ralo, A. Derevianko, A. Kropotov, S. Styervoyedov and O. Vozniy
I/O modules
Programming mechanism
Typically, these parts are combined with the use of crates on a physical level and a
number of tires on the logical one.
Despite the structural similarity of the PLC with the PC, the PLC must have some
specific characteristics. For the experiments, and automation systems building using
PLC, they should be able to work under different, sometimes very hard conditions,
ensuring high availability.
Experimental plant for the
ions and neutral particles
obtaining
Control
Sensors elements
PLC
System Storage
arbiter system
Fig. 1. The structure of the nodes communication using PLC
In the development of the PLC as a system bus selection was made from a list of
the most popular, fully standardized well studied buses. The choice was made in fa-
vour of bus VMEbus [5].
By using this bus is possible to build scalable systems, where it can be increased
both the number of modules that are responsible for input/output, and the number of
processor modules, that makes it possible to distribute the control program in the case
of its complexity.
Block diagram of the created PLC is shown in Fig. 2. As a PLC’s control processor
it has been selected 32-bit microcontroller AT91SAM7S from Atmel.
3 The Model Used
To provide a control on a given algorithm of complex technological processes, that
includes the induction plasma source management, it is necessary to develop a model
of the process. The best way to obtain it is to process the results of passive experiment
that are stored in specially designed storage that provides quick access to historical
data. After receiving the necessary amount of experimental data and the model crea-
tion, the control system can proceed in an active phase, i.e. manage the process.
Modeling, Algorithms and Implementation of the Microcontroller Control System 33
Processor part PLC interface Memory
Application memory
AT91SAM7S128 Front panel Po w e r
sou rce
Time rs L o gic
Indication
USB 2.0
and control System User Data
VME bus memory memory memory
controller
VME bus
Join panel
Input/output modules
Outer bus RS-485
Fig. 2. PLC Block diagram
To construct the control system it was decided to use the model of the 5th class by
Shantikumar [6]. In the proposed model, various simulation results for different initial
conditions are used. In addition, on the basis of these results analytical model is gen-
erated. It is possible to identify the following properties of the model:
Simulation model is used to determine the relationship between variable parame-
ters and target factor
An analytical model is a generalization of the results of various simulations; such
systems are more accurate, than the solutions obtained only by using theoretical
calculations, since they are based on multiple real data
After receiving the analytical solution, which is not changing under the new simu-
lations, this model can be used in the process control system to predict its results
for the given parameters and, hence, it provides accurate information for quick de-
cision-making [7]
The proposed class of models of simulation plays an important role in the investi-
gation of the system behaviour. The results are used to derive analytical models to
predict system performance. There may be a situation when the analytical model
without simplifications can not be created, and the approximate model is not suffi-
ciently accurate. Thus, models of this class are applicable in the cases:
When the correlation between the target factor and the system parameters is not
known that makes the analytical model is very difficult to develop, in this case,
when analytical models are very expensive, unreliable or impractical, simulation
can help in understanding the relationships between all factors that makes it possi-
ble to develop an analytical model
In many practical problems the useful signal is often superimposed with lot of
noise, which is almost impossible to take into account, simulation allows to inves-
tigate the behaviour of dynamic systems and to identify key parameters for evalua-
tion, so that these parameters are then included in the analytical model
The process of a given class model constructing was described in [8]. As a result of
simulation data processing system receives a functional relationship between the ex-
periment factors and coefficients of these relationships. To find a list of dependencies
34 A. Ralo, A. Derevianko, A. Kropotov, S. Styervoyedov and O. Vozniy
there has been used a list of 34 functions [9], that is sufficient to describe the most of
physical processes.
4 Simulation of the Control System
During the system development phase to select the most efficient architecture of its
work simulation was carried out. At the same time a number of software products
used to data networks simulating was observed.
OPNET and NetSim++ – graphical network modelling environment;
SMURPH (University of Alberta) and Ptolemy (Berkeley) use an eclectic language
for describing data lines;
OMNet++ that uses its own language to describe the architecture that is then trans-
lated by the pre-processor to standard C++.
The choice was made in favour of the last instrument, as being open, easy embed-
dable in third-party software products and having multiple trusted code in its compo-
sition.
OMNet++ is not a programming language – it’s just a class library for the simula-
tion. These classes are: modules, gateways, connections, settings, reports, histograms,
assemblers, precision registers, etc.
First of all, it was composed a system model, built using “common bus” architec-
ture with a central arbitration. The algorithm works as follows: arbiter sends a request
to the sensors. The intensity of the queries depends on the particular sensor. The re-
quest is broadcast, thus it contains the sensor’s address. Sensor, recognizing the ad-
dress, sends the data back. The size of datagram is also dependent on the sensor’s
type.
The appearance of the simulation software modelling the system is shown Figure 3
(a).
(a) (b)
Fig. 3. OMNet++ work for simplified systems
Modeling, Algorithms and Implementation of the Microcontroller Control System 35
To simulate the operation using the PLC part of the terminal devices was con-
nected to an intermediate device, which operates as a PLC. The appearance of the
resulting system in the program OMNet presented in Figure 3 (b).
Time and composition of each message sent and received is saved in a special vec-
tor file, that is then processed to obtain statistical information. As a result, basing on
multiple system runs and changing the parameters such as: number of devices, data
rate, processing time, etc. it was found that the use of PLC increases the productivity
of an average of 1.5 times. Figure 4 shows the number of posts at different loads on
the media for the cases of classical architecture and the use of the PLC. The fact that
the use of OMNet++ allows programming in C++, made it possible to produce a bust
of the parameters automatically.
Fig. 4. Dependence of the sent messages number from the bus bandwidth
5 The Results of the Developed System Application
Concrete implementation of information management system was worked out for the
experimental setup of vacuum-plasma nanostructures etching, schematically depicted
in Figure 5 [10].
The vacuum chamber is a cylindrical volume of diameter 600 mm and length
800 mm. On the side flange of the chamber it was mounted an RF ion source based on
the inductive discharge. With the ions extracted passage through the electrode system
the beam of desired particle charge or neutral particles is formed. In front of the
source a quadrupole mass spectrometer with an integrated energy analyzer with a
resolution of 0.1 eV mounted. Plasma excitation is done with an inductor connected to
the RF generator operating at a frequency of 13.56 MHz.
The management system’s task was a search for optimal etching parameters and
the stabilization of these parameters to ensure the repeatability of the experiment, as
well as to change the programmable modes of control elements.
Modes of RF discharge source and power blocs forming voltages on the beam ex-
traction system are controlled by PLC, that uses the internal bus commands to control
elements, responsible for setting the pulse’s amplitude, polarity, duration and duty
36 A. Ralo, A. Derevianko, A. Kropotov, S. Styervoyedov and O. Vozniy
cycle. PLC via USB 2.0 was connected to the computer where the accumulation of
experimental data and the search for functional relationships between them was done.
The system supports the operator control mode and independent search of optimal
parameters for this type of beams and maintains these parameters constant. The user
interface and the corresponding signals taken from the outputs of power blocks are
shown in Figure 6.
AD RFS
GIS
S
M
IND G1-3
Q
VC
- PB +
+ PB -
- PB + MMS
+ PB -
Fig. 5. Structural diagram of the setup for nanostructures etching. VC – vacuum chamber for
ion-plasma treatment, QMS – quadrupole mass spectrometer with energy analyzer, RFS – RF
source, AD – agreement device, PB – power blocks IND – inductor, G1-3 tree-grade electrode
system, GIS – gas inlet system, MMS – microcontroller management system.
Fig. 6. The user interface and the corresponding control signals
Modeling, Algorithms and Implementation of the Microcontroller Control System 37
6 Conclusion
Three-level information management system of vacuum-plasma system with the ion
source based on high-frequency induction discharge is designed to produce beams of
positive and negative ions and neutralized beam of particles used in the nanostruc-
tures etching. On the middle hierarchical level of the system there are specially de-
signed programmable logic controllers based on a 32-bit microcontroller from Atmel
AT91SAM7S, that have increased noise immunity characterized, advanced comput-
ing power and optimal capacity for a given task. PLCs make the primary information
processing and greatly accelerate the process of information exchange, particularly at
high pulse loads.
Despite the overall effectiveness of the method of constructing process models to
manage them based on statistical data accumulated, in practice, large amounts of data
to process and requirements for high-speed make this method not always usable.
Therefore, the task of modelling and process control between the main control
mechanism and the PLC allocation is important.
Testing of the system to manage the real process has shown its efficiency and al-
lowed to obtain an adequate empirical model of the controlled ion source based on
high-frequency discharge from tree-grades control.
References
1. Kinoshita, T., Hane, M., McVittee, J. P.: Journal of Vacuum Science and Technology
B 14, 560 (1996)
2. Yunogami, T., Yokogawa, K., Mizutani, T.: Development of neutral-beam-assisted etcher.
Journal of Vacuum Science and Technology A 13(3), 952 (1995)
3. Yokogawa, K., Yunogami, T., Mizutani, T.: Neutral- Beam-Assisted Etching System for
Low-Damage SiO2 Etching of 8-Inch Wafers. Japan Journal of Application Physics 35,
1901 (1996)
4. Vozniy, O. V., Yeom, G. Y.: High-energy negative ion beam obtained from pulsed induc-
tively coupled plasma for charge-free etching process. Appl. Phys. Lett. 94, 231502 (2009)
5. VITA. Open standards, open markets, http://www.vita.com (2009)
6. Shanthikumar, J. G., Sargent, R. G.: Unifying view of hybrid simulation/analytic models
and modeling. Operations Research 31 (6), 1030-1052 (1983)
7. Hsieh, T.: Hybrid analytic and simulation models for assembly line design and production
planning. Simulation Modeling Practice and Theory 10, pp. 87-108 (2002)
8. Derevianko, A. V.: Constructing empirical models for the management of complex techno-
logical processes. Bulletin of V.N. Karazin Kharkiv National University. No. 12: Mathe-
matical modeling. Information technology. Automated control systems. № 863 - Kharkov:
Publishing House of the KNU, P.101-110 (2009) (In Russian)
9. Kuri-Morales, A., Rodriguez-Erazo, F.: A search space reduction methodology for data
mining in large databases. Engineering Application of Artificial Intelligence 22, pp. 57-65
(2009)
10. Vozniy, O., Polozhiy, K., Yeom, G. Y.: Journal of Application Physics 102 083306 (2007)
Using Algebra-Algorithmic and Term Rewriting Tools
for Developing Efficient Parallel Programs
Anatoliy Doroshenko1, Kostiantyn Zhereb1 and Olena Yatsenko1
1 Institute of Software Systems of National Academy of Sciences of Ukraine,
Glushkov prosp. 40, 03187 Kyiv, Ukraine
doroshenkoanatoliy2@gmail.com, zhereb@gmail.com, oayat@ukr.net
Abstract. An approach to program design and synthesis using algebra-
algorithmic specifications and rewriting rules techniques is proposed. An alge-
bra-algorithmic toolkit based on the approach allows building syntactically cor-
rect and easy-to-understand algorithm specifications. The term rewriting system
supplements the algebra-algorithmic toolkit with facilities for transformation of
the sequential and parallel algorithms, enabling their improvement.
Keywords. Algebra of algorithms, code generation, formalized design of pro-
grams, parallel computation, term rewriting
Key terms. FormalMethod, HighPerformanceComputing, ConcurrentComputa-
tion, Integration
1 Introduction
Nowadays uniprocessor systems are almost fully forced out by multiprocessor ones,
as the latter allow getting the considerable increase of productivity of programs. Thus,
the need of program parallelization arises [10]. There are libraries, such as pthreads,
OpenMP, TBB and others [1], allowing developers to write parallel programs. Using
these libraries a programmer manually divides code into independent sections, de-
scribes data exchange and synchronization between them. However, such method has
substantial defects, in particular, related to committing of errors into program code
and a time required for parallelization and debugging. Therefore, the parallelization
process has to be automatized as much as possible, and in an ideal, should be carried
out fully automatically, without participation of a programmer.
This paper continues our research on automation of process of designing and de-
velopment of efficient parallel programs, started in [2], [9], [10], [11]. Our approach
is based on usage of Integrated toolkit for Designing and Synthesis of programs (IDS)
[2], [19]. The process of algorithm designing in IDS consists in the composition of
reusable algorithmic components (language operations, basic operators and predi-
cates), represented in Systems of Algorithmic Algebras (SAA) [2], [9], [19]. We used
IDS for generation of sequential and parallel programs in Java and C++ on the basis
Using Algebra-Algorithmic and Term Rewriting Tools … 39
of high-level algorithm specifications (schemes). To automate the transformations of
algorithms and programs we use term rewriting system Termware [8], [11]. The nov-
elty of this paper is 1) adjusting IDS to generate parallel code in Cilk++ language,
which is an extension to the C and C++ programming languages, designed for multi-
threaded parallel computing [7] and 2) closer integration between IDS and Termware
systems. The approach is illustrated on a recursive sorting algorithm (quick sort).
The problem of automated synthesis of program code from specifications has been
studied extensively and many approaches have been proposed [13], [14]. Important
aspects of program synthesis include 1) format of inputs (specifications), 2) methods
for supporting concrete subject domains and 3) techniques for implementing trans-
formation from specifications to output program code (these aspects roughly corre-
spond to 3 dimensions of program synthesis discussed in [14]). For input specifica-
tion, a popular option is using domain-specific languages (DSLs) [4], [17] that allow
capturing requirements of subject domain. Other options include graphical modeling
languages [5], [17], formal specification languages [16], ontologies [6] and algebraic
specifications [3]. Using such formalisms enables analysis and verification of specifi-
cations and generated code. There are also approaches that provide specification not
of program or algorithm, but of problem to be solved, in form of functional and non-
functional constraints [18], examples of input/output pairs [15], or natural language
descriptions [14].
Another crucial aspect of program synthesis is specialization for subject domain.
Some approaches are restricted to a single domain, such as statistical data analysis
[12] or mobile application development [17]; others provide facilities for changing
domain-specific parts, by using ontological descriptions [6], grammars [16], or by
providing generic framework that is complemented by domain-specific tools [18].
Finally, an important aspect is transformation from input specification into source
code in a target language. A transformation algorithm can be hand-coded [12], but it
reduces flexibility of system. Therefore, transformation is often described in a de-
clarative form, such as rewriting rules [16], visualized graph transformations [17],
code templates [6]. More complex approaches require searching the space of possible
programs [18], possibly using genetic programming or machine learning approaches
[14]. In [4], partial synthesis is proposed: generic parts of application are generated,
and then completed with specific details manually.
In comparison, our approach uses algebraic specifications, based on Glushkov al-
gebra of algorithms [2], but they can be represented in three equivalent forms: alge-
braic (formal language), natural-linguistic and graphical, therefore simplifying under-
standing of specifications and facilitating achievement of demanded program quality.
Another advantage of IDS is a method of interactive design of syntactically correct
algorithm specifications [2], [19], which eliminates syntax errors during construction
of algorithm schemes. Specialization for subject domain is done by describing basic
operators and predicates from this domain. Our approach uses code templates to spec-
ify implementations for operators and predicates; program transformations, such as
from sequential to parallel algorithm, are implemented as rewriting rules. Such sepa-
ration simplifies changing subject domain or transformations.
40 A. Doroshenko, K. Zhereb and O. Yatsenko
2 Formalized Design of Programs in IDS and Termware
The developed IDS toolkit is based on System of Algorithmic Algebras (SAA), which
are used for formalized representation of algorithmic knowledge in a selected subject
domain [2], [9], [19]. SAA is the two-based algebra SAA = <{U, B}; >, where U is a
set of logical conditions (predicates) and B is a set of operators, defined on an infor-
mational set; = 1 2 is the signature of operations consisting of the systems 1
and 2 of logical operations and operators respectively (these will be considered be-
low). Operator representations of algorithms in SAA are called regular schemes. The
algorithmic language SAA/1 [2] is based on mentioned algebra and is used to describe
algorithms in a natural language form. The algorithms, represented in SAA/1, are
called SAA schemes.
Operators and predicates can be basic or compound. The basic operator (predicate)
is an operator (predicate), which is considered in SAA schemes as primary atomic
abstraction. Compound operators are built from elementary ones by means of opera-
tions of sequential and parallel execution operators, branching and loops, and syn-
chronizer WAIT ‘condition’ that delays the computation until the value of the
condition is true (see also Table 1 in next section).
The advantage of using SAA schemes is the ability to describe algorithms in an
easy-to-understand form facilitating achievement of demanded quality of programs.
The IDS is intended for the interactive designing of schemes of algorithms in SAA
and generating programs in target programming languages (Java, С++, Cilk++). In
IDS algorithms are designed as syntactically correct programs ensuring the syntactical
regularity of schemes. IDS integrates three forms of design-time representation of
algorithms: regular schemes, SAA schemes (textual representation of SAA formulae)
and flow graphs. For integration with Termware, in this paper IDS was also adjusted
on generation of programs in Termware language.
Fig. 1. Architecture of the IDS toolkit
The IDS toolkit consists of the following components (Fig. 1): constructor, in-
tended for dialogue designing of syntactically correct sequential and concurrent algo-
rithm schemes and generation of programs; flow graph editor; generator of SAA
schemes on the basis of higher level schemes, called hyper-schemes [19]; and data-
Using Algebra-Algorithmic and Term Rewriting Tools … 41
base, containing the description of SAA operations, basic operators and predicates in
three mentioned forms, and also their program implementations.
The constructor is intended to unfold designing of algorithm schemes by superpo-
sition of SAA language constructs, which a user chooses from a list of reusable com-
ponents for construction of algorithms. The design process is represented by a tree of
an algorithm [2], [19]. On each step of the design process the constructor allows the
user to select only those operations, the insertion of which into the algorithm tree does
not break the syntactical correctness of the scheme. The tree of algorithm constructing
is then used for automatic generation of the text of SAA scheme, flow graph and the
program code in a target programming language.
Example 1. We illustrate the use of SAA on Quicksort algorithm, which is given
below in the form of SAA scheme. The identifiers of basic operators in the SAA
scheme are written with double quotes and basic predicates are written with single
quotes. Notice that identifiers can contain any text explaining the meaning of operator
or predicate. It is not interpreted: it has to match exactly the specification in the data-
base (however, since constructs are not entered manually, but selected from a list, the
misspellings are prevented). The comments and implementations of compound opera-
tors and predicates in SAA schemes begin with a string of “=” characters.
SCHEME QUICKSORT_SEQUENTIAL ====
"main(n)"
==== Locals (
"Declare an array (a) of type (int) and size (n)";
"Declare a variable (i) of type (int)";
"Declare a variable (end) of type (int)");
"Fill the array (a) of size (n) with random
values";
"end := a + n";
"qsort(a, end)";
"qsort(begin, end)"
==== IF NOT('begin = end')
"Reduce (end) by (1)";
"Reorder array (a) with range (begin) and (end)
so that elements less than pivot (end) come
before it and greater ones come after it; save
pivot position to variable (middle)";
"qsort(begin, middle)";
"Increase (middle) by (1)";
"Increase (end) by (1)";
"qsort(middle, end)"
END IF
END OF SCHEME QUICKSORT_SEQUENTIAL
42 A. Doroshenko, K. Zhereb and O. Yatsenko
To automate the transformation (e.g. parallelization) of programs we augment ca-
pabilities of IDS with rewriting rules technique [8], [11]. At the first step we construct
high-level algebraic models of algorithms based on SAA in IDS (see also [2], [9],
[19]). After high-level program model is created, we use parallelizing transformations
to implement a parallel version of the program on a given platform (multicore in this
paper). Transformations are represented as rewriting rules and therefore can be ap-
plied in automated manner. The declarative nature of rewriting technique simplifies
adding new transformations. Also transformations are separated from language defini-
tions (unlike approach used in [16]), therefore simplifying addition of new transfor-
mations or new languages.
We use the rewriting rules system Termware [8], [11]. Termware is used to de-
scribe transformations of terms, i.e. expressions in a form f t1 , , tn . Transforma-
tions are described as Termware rules, i.e. expressions of form source [condi-
tion]-> destination [action]. Here source is a source term (a pattern
for match), condition is a condition of rule application, destination is a trans-
formed term, action is additional action that is performed when rule fires. Each of
4 components can contain variables (denoted as $var), so that rules are more gener-
ally applicable. Components condition and action are optional. They can exe-
cute any procedural code, in particular use the additional data on the program.
3 Generation of Terms and Programs and Experimental Results
IDS system performs generation of programming code on the basis of an algorithm
tree, received as a result of designing an algorithm in the IDS Constructor (see Sec-
tion 2), and also code templates – implementations of basic operators and predicates
in a target language (Java, C++, Cilk++), that are stored in IDS database. In the proc-
ess of generation, IDS translates SAA operations into corresponding operators of
programming language. Compound operators can be represented as subroutines
(methods). IDS database contains various code patterns for generation of parallel
programs, namely using WinAPI threads, Message Passing Interface (MPI), and
Cilk++ operations [7]. For implementation of parallel version of our illustrative ex-
ample (Quicksort algorithm), we used Cilk++ as it facilitates programming of recur-
sive parallel programs [7]. Cilk++ is a general-purpose programming language, based
on C/C++ and designed for multithreaded parallel computing.
Table 1 gives a list of main SAA operations and templates of their implementation
in Termware and Cilk++, which are stored in the IDS database. The implementations
contain placeholders like ^condition1^, ^operator1^ etc., which are replaced
with program code during the program generation.
For the purpose of transformation of some algorithm, IDS performs the generation
of a corresponding term and developer specifies a set of rules for transformation.
Then Termware carries out the actual transformation, the result of which can further
be used for code generation in a programming language.
Using Algebra-Algorithmic and Term Rewriting Tools … 43
Table 1. The main SAA operations and templates of their implementation in Termware and
Cilk++ languages
Text of SAA operation Termware implementation Cilk++ implementation
“operator1”; then (^operator1^, ^operator1^;
“operator2” ^operator2^) ^operator2^
IF ‘condition’ IF (^condition1^, if (^condition1^){
THEN “operator1” ^operator1^, ^operator1^ }
ELSE “operator2” ELSE (^operator2^)) else {^operator2^}
END IF
FOR '(var) from FOR (%1, %2, %3, for (%1, %2, %3) {
(begin) to ^operator1^ ^operator1^
(end)' ) }
LOOP “operator1”
END OF LOOP
(“operator1” Parallel( cilk_spawn
PARALLEL “opera- ^operator1^, ^operator1^;
tor2”) ^operator2^) ^operator2^
WAIT ‘condition’ WAIT cilk_sync;
(^condition1^)
Example 2. We will parallelize the sequential Quicksort algorithm (see Exam-
ple 1), using IDS and Termware. For the parallelization, function qsort has to be
transformed, so we generated the term for this function:
qsort(Params(begin, end),
IF (NOT(Equal(begin, end)),
then (Dec(end, 1),
then (Partition(a, begin, end, end),
then (CALL(qsort(begin, middle)),
then (Inc(middle, 1),
then (Inc(end, 1),
CALL (qsort(middle, end)))))))))
Then the operation of parallel execution of operations has to be added to this term.
This is done by applying the following two Termware rules:
1. then(CALL($x), then ($y, $z)) ->
Parallel (CALL($x), then($y, $z))
2. then($x1, Parallel($x2, $x3)) ->
then($x1, then(Parallel($x2, $x3),
WAIT(AllThreadsCompleted(n))))
The first rule replaces the operation of sequential execution of operators with paral-
lel execution. The second rule adds a synchronizer WAIT(AllThreads Com-
44 A. Doroshenko, K. Zhereb and O. Yatsenko
pleted(n), which delays the computation until all threads complete their work.
The result of the transformation is given below.
qsort(Params(begin, end),
IF(NOT(Equal(begin, end)),
then (Dec(end, 1),
then (Partition(a, begin, end, end),
then (Parallel(
CALL (qsort(begin, middle)),
then (Inc(middle, 1),
then (Inc(end, 1),
CALL (qsort(middle, end))))),
WAIT(AllThreadsCompleted(n)))))))
Thus, as a result of parallelization, the first operator (thread) of Parallel opera-
tion executes the operator qsort(begin, middle), and the second one calls
two Inc operators and qsort(middle, end). Operation
WAIT(AllThreadsCompleted(n)) performs the synchronization of
threads. The threads are created recursively; their quantity is specified as an input
parameter of function main. Notice that these transformations are only valid if two
qsort calls are independent. The system doesn’t check this property: it has to be
asserted by a developer.
The resulting parallel algorithm scheme Quicksort was used for generation of code
in Cilk++ using IDS system. The parallel program was executed on Intel Core 2 Quad
CPU, 2.51 GHz, Windows XP machine. Fig. 2 shows the program execution time in
seconds. The speedup at execution of program with usage of 2, 3 and 4 processors
was 2; 2.9 and 3.8 accordingly, which shows that the program has a good degree of
parallelism and is scalable.
Fig. 2. The execution time of parallel Quicksort program on a quad-core processor; the size of
input array is 5107 elements
Using Algebra-Algorithmic and Term Rewriting Tools … 45
4 Conclusion
We have described our approach of constructing efficient parallel programs using
high-level algebra-algorithmic specifications and rewriting rules technique. Algebra-
algorithmic toolkit IDS and rewriting rules engine Termware are combined to enable
formal, yet easy-to-understand algorithm specifications and automate program syn-
thesis and parallelization process. The combined development toolkit can be retar-
geted to various subject domains and implementation languages, as exemplified by
Cilk++. The developed system could be further extended with automated code analy-
sis facilities based on rewriting technique.
References
1. Akhter, S., Roberts, J.: Multi-Core Programming. Intel Press, Hillsboro (2006)
2. Andon, F. I., Doroshenko, A. Y., Tseytlin, G. O., Yatsenko, O. A.: Algebra-Algorithmic
Models and Methods of Parallel Programming. Akademperiodika, Kyiv (2007) (in Rus-
sian)
3. Apel, S. et al.: An Algebraic Foundation for Automatic Feature-Based Program Synthesis.
Science of Computer Programming. 75(11), 1022–1047 (2010)
4. Bagheri, H., Sullivan, K.: Pol: Specification-Driven Synthesis of Architectural Code
Frameworks for Platform-Based Applications. In: Proc. 11th Int. Conf on Generative Pro-
gramming and Component Engineering, pp. 93–102, ACM, New York (2012)
5. Batory, D.: Program Refactoring, Program Synthesis, and Model-Driven Development. In:
Proc. 16th Int. Conf. on Compiler Construction. LNCS 4420, pp. 156–171 Springer-
Verlag, Berlin Heidelberg (2007)
6. Bures, T. et al.: The Role of Ontologies in Schema-Based Program Synthesis. In: Proc.
Workshop on Ontologies as Software Engineering Artifacts, Vancouver (2004)
7. Cilk Home Page, http://cilkplus.org/
8. Doroshenko A., Shevchenko R.: A Rewriting Framework for Rule-Based Programming
Dynamic Applications, Fundamenta Informaticae, 72(1–3), 95–108 (2006)
9. Doroshenko, A., Tseytlin, G., Yatsenko, O., Zachariya, L.: A Theory of Clones and For-
malized Design of Programs. In: Proc. Int. Workshop on Concurrency, Specification and
Programming (CS&P’2006), pp. 328–339, Wandlitz, Germany (2006)
10. Doroshenko, A. Y., Zhereb, K. A., Yatsenko, Ye. A.: On Complexity and Coordination of
Computation in Multithreaded Programs. Problems in Programming, 2, 41–55 (2007) (in
Russian)
11. Doroshenko, A., Zhereb, K.: Parallelizing Legacy Fortran Programs Using Rewriting
Rules Technique and Algebraic Program Models. In: Ermolayev, V. et al. (eds.) ICT in
Education, Research, and Industrial Applications. CCIS 347, pp. 39–59. Springer Verlag,
Berlin Heidelberg (2013)
12. Fischer, B., Schumann, J.: AutoBayes: a System for Generating Data Analysis Programs
from Statistical Models. J. Funct. Program. 13(3), 483–508 (2003)
13. Flener, P.: Achievements and Prospects of Program Synthesis. In: Kakas, A. C., Sadri, F.
(eds.) Computational Logic: Logic Programming and Beyond. LNCS 2407, pp. 310–346,
Springer Verlag, London (2002)
46 A. Doroshenko, K. Zhereb and O. Yatsenko
14. Gulwani, S.: Dimensions in Program Synthesis. In: 12th Int. ACM SIGPLAN Symposium
on Principles and Practice of Declarative Programming, pp. 13–24. ACM, New York
(2010)
15. Kitzelmann, E.: Inductive Programming: a Survey of Program Synthesis Techniques. Ap-
proaches and Applications of Inductive Programming, LNCS 5812, pp. 50–73. Springer
Verlag, Berlin Heidelberg (2010)
16. Leonard, E. I., Heitmeyer, C. L.: Automatic Program Generation from Formal Specifica-
tions using APTS. In: Automatic Program Development. A Tribute to Robert Paige, pp.
93–113. Springer Science, Dordrecht (2008)
17. Mannadiar, R., Vangheluwe, H.: Modular Synthesis of Mobile Device Applications from
Domain-Specific Models. In: Proc. 7th Int. Workshop on Model-Based Methodologies for
Pervasive and Embedded Software, pp. 21–28. ACM, New York (2010)
18. Srivastava, S., Gulwani, S., Foster, J. S.: From Program Verification to Program Synthesis.
In: Proc. 37th Annual ACM SIGPLAN-SIGACT Symposium on Principles of Program-
ming Languages, pp. 313–326. ACM, New York (2010)
19. Yatsenko, O.: On Parameter-Driven Generation of Algorithm Schemes. In: Proc. Int.
Workshop on Concurrency: Specification and Programming (CS&P’2012), pp. 428–438,
Berlin, Germany (2012)
1.2 Machine Intelligence, Knowledge
Engineering and Management for ICT
An Intelligent Approach to Increase Efficiency of IT-
Service Management Systems: University Case-Study
Nikolay Tkachuk1, Vladyslav Sokol1 and Kateryna Glukhovtsova1
1
National Technical University “Kharkov Polytechnic Institute”
Frunze str., 21, Kharkov, Ukraine
tka@kpi.kharkov.ua, vladislav.sokol@gmail.com, kat_1109@mail.ru
Abstract. A comprehensive framework to increase efficiency of IT-services
management systems (ITSMS) is proposed, which resolves 3 interconnected
tasks in a target organization: 1) providing an effective configuration of ITSM-
modules according to their specific features and needs; 2) integration a given
ITSMS with existing enterprise architecture; 3) advanced incidents management
in ITSMS. The applicability of this approach was tested successfully on the
case-study at the National Technical University “Kharkov Polytechnic Institute”
(www.kpi.kharkov.ua ).
Keywords. IT-service management, effectiveness, multi-criteria ranking, data
integration, adaptive ontology, case-based reasoning, e-learning
Key terms. Academia, ICTInfrastructure, KnowledgeManagementProcess,
Model, SoftwareEngineeringProcess
1 Introduction: Problem Actuality and Research Objectives
Nowadays the concept of ITIL (IT Infrastructure Library) [1] and the new kind of
computerized management systems, namely: IT Service Management Systems
(ITSMS) became a growing and perspective approach to solve very important and
complex technical problem and, at the same time, business-focused one: how to or-
ganize a well-structured and controllable IT-environment at an appropriate organiza-
tion?
According to ISO/IEC 20000 [2] an IT Service Management System (ITSMS) pro-
vides “…a framework to enable the effective management and implementation of all
IT-services”. Due to high complex and multi-dimensional nature of IT-services in
large modern business organizations, which ITSMS are dealing with, recent publica-
tions in these domain present some sophisticated approaches to design and to use
these facilities. One such important topic in ITIL-ITSM domain is the integration of
ITSMS functionality into enterprise architecture (see, e.g. in [3,4]). Another recent
trend in ITSMS-development is the usage of ontologies and model-driven architecture
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 49
(MDA) [5, 6] for knowledge handling and re-using. Their authors emphasize the ac-
tual need to elaborate and to apply several knowledge-oriented approaches to re-
quirements analysis within ITSMS-development, and to quantitative quality assess-
ment of appropriate project solutions.
Taking into account some ITSMS-issues mentioned above, the main objective of
the research presented in this paper is to propose the first vision for intelligent com-
plex approach to increase efficiency of typical ITSMS, with a proof of concept basing
on the ITSMS university case-study. The rest of this paper is organized in the follow-
ing way: Section 2 analyses some existing ITSMS, introduces our vision about their
typical functionality, and shows the list of prioritized tasks to be resolved to increase
an efficiency of an ITSMS. In Section 3 we present the method elaborated for effec-
tive ITSM-modules configuring in a target business organization, and Section 4 re-
ports the first version of ITSMS - ontologies to integrate the selected modules into
enterprise architecture (EA). In Section 5 the designing perspective for the combina-
tion case-based reasoning (CBR) with ontology-based approach to advanced incident
management it ITSMS is briefly outlined. In Section 6 we present the university case-
study for our method to estimate an effectiveness of different ITSMS configurations
and discuss the results achieved. In Section 7 the paper concludes with a short sum-
mary and with an outlook on the next steps to be done in the proposed development
framework.
2 Typical Functionality of ITSMS and the Complex of
Intelligent Tasks to Increase its Efficiency
In order to elaborate a way how to provide a complex approach to increase an effi-
ciency of ITSM-system operating, it is necessary to understand its typical functional-
ity and to analyze its specific features.
2.1 Overview of existing ITSMS
We have analyzed some already existing ITSMS [7-10], and the results of this study
is presented in the Table 1. Basically, all such systems can be divided into 3 groups,
namely: (a) advanced business ITSM-products; (b) open source ITSM-solutions; (c)
bespoke ITSM-systems.
To the group (a) belong such systems as, e.g., HP OpenView Service Desk [7] and
BMC Remedy [8]. The first software product is the absolutely leader in this market
segment, because the most part of organizations which prefer ITSM-business solu-
tions from the group (a), are using exactly HP-platform. The number of its running
installations is essentially less than HP, at least because of more expensive costs of
Remedy ITSM Suite.
50 N. Tkachuk, V. Sokol and K.Glukhovtsova
Table 1. Results of comparison for some ITSMS
OMNINET
HP Service
BMC Remedy Axios OmniTracker
Criteria / Systems Manager
ITSM Suite 7.5 Assyst 7.5 ITSM Center
7.10
2.0
Basic
5 5 5 4
functionality
Maintainability 5 4 5 4
Report generation 4 5 5 4
Scaleability 4 2 3 5
Web-interface 5 5 5 5
ITSM-solutions from the group (b) also are used in practice, but they definitely
have limited functionality and provide less level of IT-services management. The
typical open source ITSMS are, for instance, GLPI [8], OTRS [9], and some others,
which are listed at the Web-resource SourceForge [10].
And, objectively, the business organizations, which are not ready to buy advanced
software products from group (a), and which are not satisfied with functionality pro-
vided by ITSM-systems from group (b), because they have some specific IT-needs
and challenges, exactly these companies try to develop their own ITSM-solutions to
be considered as members of the group (c). The more detailed comprehensive study
of some existing ITSM-systems is presented in [11].
2.2 Typical ITSMS-functionality
Based on the given analysis of the real ITSMS (see above), we have elaborated the
following vision for their typical functionality (see Fig. 1).
There are 5 main subsystems (or packages) of system functions, namely:
1. IT Business Alignment: this subsystem is supposed to implement a ІТ-strategy in
given business organization with respect to its main goals and needs, and to pro-
vide a base for costs assessment to whole IT-infrastructure;
2. Service Operations: this facility is responsible for customer’s requests management
(regarding to a current incident and to a related problem), and for providing of
ITSM-support functions;
3. Service Delivery Assurance: this functional package implements a configuration
and change management of all ITSM-software tools thus is extremely important
for a stable IT-environment;
4. Service Design and Management: this ITSMS-functionality provides detailed in-
formation about new perspective IT-services to be designed with respect to their
availability and quality for IT-customers;
5. Service Development and Deployment: this subsystem allows to create and to test
new ITSM-services and appropriate IT-infrastructure solutions, including installa-
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 51
tion of new hard-ware components, development of additional software applica-
tions, and training programs for ITSM-staff’ and for end-users as well.
As we can see on the structure presented in Fig.1, each of these 5 subsystems is
built from several functional modules (they are depicted as UML-classes). The most
important of them are the following ones:
Module М1 =”Incident management”: it includes organizational procedures and
appropriate tools to resolve current incidents, which IT-service users are facing with
(hard-and software errors, network connection problems, request for consultations,
etc.);
Fig. 1. Typical functionality of a ITSMS
Module М2 = ”Problem management”: this facility provides tools to detect and to
eliminate any problem situation which is a reason for different incidents;
Module М3 = ”Configuration management”: this module supports all operating
sub-schemes in the IT-infrastructure of given business organization;
Module М4 = ”Change management”: it supervises and coordinates all changes
which arise in IT-infrastructure; .
Module М5 =”Service level management”: this unit is responsible for definition
and implementation of an appropriate level of IT-services to be provided for cus-
tomers.
In ITIL-best practice manuals (see e.g. in [12]) the following 3 main schemes are
considered to introduce these modules into IT-infrastructure of a target organization: a
classic scheme (S1); a contract scheme (S2); an infrastructure-centered scheme (S3).
52 N. Tkachuk, V. Sokol and K.Glukhovtsova
A classic scheme S1 is the most applied solution in the ITSM-domain, and it supposes
the following sequence of modules М1-М5:
S1 M 1, M 3, M 4, M 2, M 5 (1)
This approach quickly allows to resolve the most actual communication problems
between IT-service department and customers basing on incident management (mod-
ule М1), and it provides some tools for all IT-services support (the modules М3 and
М4), and after that a platform for future IT-infrastructure development is introduced
(modules М2 та М5 respectively). But in this case it has to be taken into account this
scheme is a most expensive way for a given business organization, and it requires a
lot of resources exactly at an initial phase of whole ITSM-configuring framework.
A contract scheme S2 actually aims to formalize a communication process between
IT-service department and customers, and it has the following modules-workflow:
S 2 M 5, M 3, M 1, M 4, M 2 (2)
In this case all customer requirements to IT-services have to be collected and
specified (in module М5), and appropriate IT-infrastructure sub-schemes can be built
(using module М3), in order to define prospective IT-strategy in the target organiza-
tion, next an operative ITSM-functionality is provided, including incident manage-
ment (in module М1), change management (in module М4), and problem manage-
ment (in module М2). Obviously, this scheme definitely has some risk factors regard-
ing its efficiency, if the initial IT-service specifications were done not correctly (in
module М5).
And, finally, an infrastructure-centered scheme S3 proposes the modules sequence
indicated as following:
S 3 M 3, M 4, M 2, M 1, M 5 (3)
that is, firstly, to provide tools for all IT-services support (modules М3 and М4 re-
spectively). Secondly, this approach allows to manage all typical problem situations
(in module М2), and already based on this one to detect and to resolve corresponded
incidents by IT-service customers (in module М1). Thirdly, it creates an opportunity
to define in computer-aided way the necessary composition and the IT-service level
management (in module М5).
It is necessary to note that besides some empirical recommendations concerning
the possible ITSM-modules configurations defined as (1)-(3), in the appropriate tech-
nical documentation there are no more or less proved suggestions about possible
quantitative estimations for effectiveness of these alternative approaches.
2.3 The complex of intelligent tasks to increase of ITSM-system efficiency
Taking into account the results of performed analysis (see above), and based on some
modern trends in the domain of ITSMS-development (see Section 1), the following
list of prioritized tasks can be composed in order to increase ITSMS-efficiency,
namely
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 53
1. to provide an effective configuring of ITSM-modules for a target organization,
taking into account its specific features and needs;
2. to elaborate an integration framework for a given ITSM-system’s configuration
and for an existing enterprise architecture (EA);
3. to support an advanced incidents management in the already installed ITSM-
system.
In our opinion, the task (I) can be resolved basing on some expert methods for
multi-criteria ranking, with respect to specific IT-infrastructure’s features and cus-
tomer needs in a concerned business organization [13,14]. The task (II) belongs to
already well-known integration issues in distributed heterogeneous information sys-
tems, and e.g. an ontology-based approach can be used for this purpose (e.g. in
[3,6,15]). And, finally, to solve the task (III) an additional decision-making function-
ality for typical ITSM-services (see Fig.1) has to be elaborated, e.g. basing on the
combination of case-based reasoning (CBR) approach with ontologies [16,17]. Below
these tasks and their possible solutions are presented and discussed in more detail.
3 The Method for Effectiveness Estimation of Alternative
ITSM-Module Configurations
To formalize the task (I) from their list considered in the Section 2.3, namely: to
provide an effective configuring of ITSM-modules for a target business organization,
the following factors have to taken into account: such a problem has a high
complexity grade and it is semi-formalized; estimation criteria for it are of different
nature and they are multi-valued; an information base to solve this task mainly can be
collected basing on expert data only; available expert data could be quantitative and
qualitative values both.
To solve this task we have chosen one of the multi-criteria ranking methods,
which is presented in [14]. Accordingly to this approach the following steps have to
be performed:
Step 1. A set of possible alternatives,
X { x1 , x 2 ,..., x n } { x i , i 1, n} (4)
and a set of global importance criteria to characterize these alternatives
K {K1, K 2 ,..., K m } {K j , j 1, m} (5)
have to be defined.
Step 2. Each global criteria K j is characterized by a subset of appropriate local crite-
ria
K j {k j1 , k j 2 ,..., k jQ } {k jq , q 1, Q} (6)
further, a set of membership functions according to all local criteria alternatives
54 N. Tkachuk, V. Sokol and K.Glukhovtsova
{ k j1 ( xi ), k j 2 ( xi ),..., k jQ ( xi )} { k jq ( xi ), q 1, Q, j 1, m} (7)
and the weight coefficients of their relative importance for these local criteria
{w j1 , w j 2 ,..., w jQ } {w jq , q 1, Q} (8)
have to be determined, where the following condition has to be fulfilled
Q
w jq 1 (9)
q 1
Step 3. To determine membership functions of alternatives { x i , i 1, n} to criteria
K j ,{ j 1, m} based on an additive convolution of their local criteria
Q (10)
k ( xi ) w jq k ( xi )
j jq
q 1
Table 2. Definition of membership functions for criteria to alternatives (fragment)
Alternatives Criteria K
K1 … KM
k11 … k1Q … kM 1 … k Mm
X x1 k11 ( x1 )
… … k1Q ( x1 ) … k M 1 ( x1 ) Mm ( x1 )
… … … … … … … …
xn k11 ( xn ) … k1Q ( xn ) … k M 1 ( xn ) … Mm ( xn )
… … …
Step 4. Taking into account the membership functions obtained { K j ( xi ), j 1, m} for
all alternatives xi ,{i 1, n} it is possible to determine a joined membership function for
a generalized criterion K :
m
K ( xi ) w j K j ( xi )
(11)
j 1
where w j , j 1, m are coefficients of their relative importance K j , j 1, m .
Step 5. Finally, an alternative with a maximum value of membership function for
generalized criterion K can be chosen as a target solution:
( x* ) max{ K ( xi ), i 1, n}) (12)
Below in Section 6 we present the case-study, which was performed to prove this
method, and we discuss the results achieved.
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 55
4 Ontological Specifications for ITSMS-EA Integration
Framework
As already mentioned above (see Section 2), any ITSMS has to be integrated into an
existing EA of a target organization. In our approach this task (II) has to be resolved
for an ITSMS-configuration defined with the method presented in Section 3.
This issue is already discussed intensively in a lot of publications, and their authors
consider both its conceptual and technological aspects. E.g., an ITSMS-EA integra-
tion based on well-known SOA – framework is presented in [3], and as the important
conceptual input for this issue the appropriate meta-model (actually, some kind of a
domain ontology) for IT services is designed. In [5] an approach to integration of
ITSM-services and business processes in given organization is elaborated, using onto-
logical specifications to formalize the good practice guidance for ITSM. An ontology-
based framework to integration of software development and ITSMS-functioning is
proposed in [15], thus resulting in enhanced semantic-aware support tools for both
processes. Even this brief overview allows us to conclude that exactly an ontology-
based approach is a most effective way to solve this problem. That is why, in our
opinion, to provide ITSM-EA integration effectively, it is necessary to combine the
following information resources (IR), namely: a) IR related to ITSMS – functionality,
b) IR concerned EA-domain, c) IR characterized a target organization (TO), which is
facing an ITSMS-EA integration problem with. Let’s define these IR (a)-(c) as: Onto-
ITSMS, Onto-EA, and Onto-TO respectively. Thus, the IR needed to provide an
ITSMS-EA integration should be specified using an appropriate joined ontology,
designated as Onto_ITSMS-EA.
Onto _ ITSM EA Onto ITSMS , Onto EA, Onto TO (13)
Obviously, some already existing ITIL / ITSM ontological specifications can be
used for this purpose, e.g.: Onto-ITIL ontology elaborated in [5] basing on OpenCyc
ontology (www.opencyc.org), Onto-SPEM (Software Process Engineering Meta-
model) ontology [18], and Onto-WF (WorkFlow) ontology [19]. Taking these re-
sources into account, we can represent the ontological specification for Onto_ITSM in
the following way
Onto ITSMS Onto ITIL, Onto SPEM , Onto WF (14)
There are also several ontologies developed to specify EA, and according to one of
recent and comprechensive researches in this domain presented in [20], we accept the
following 3-level definition for EA-ontology
Onto EA Onto BT , Onto AC , Onto RS (15)
where: Onto_BT is a sub-ontology of Business Terms (BT), Onto-SC is a sub-
ontology of Architecture Components (AC), and Onto-RS as a sub-ontology of
RelationShips (RS) among items of AC.
And finally, to define an Onto-TO ontology for target organization given in
expression (13), its specific features and needs related to ITSMS-usage within
56 N. Tkachuk, V. Sokol and K.Glukhovtsova
existing EA have to be taken into account. As a small excerpt of such domain-specific
Onto-TO, which is elaborated in our University-ITSMS case-study (see Section 6),
the following UML-class diagram in Fig. 2 is shown.
Fig. 2. Taxonomy of customers in a University-ITSMS as a part of a Onto-TO ontology
The proposed ontology-based approach for ITSMS-EA integration can also be used
to elaborate the solution for the task (III) from their list completed in Section 2.3.
5 Adaptive Onto-CBR Approach to Advanced Incidents
Management in ITSM
In order to solve the task (III), namely: to provide an advanced incidents management
in ITSMS, accordingly to our inter-disciplinary vision about the ITSMS-development
in general, we propose to amalgamate the following design-principles (i)-(iv) listed
below
(i) an incident management as a weak-formalized and complex task within the
ITSMS-support for its customers can effective be resolved using one of the intel-
ligent decision-support methods, e.g., using CBR-method;
(ii) to enhance a CBR-functionality, especially with respect to specific needs in a
target organization, an appropriate domain-ontology should be elaborated and
used combining with CBR;
(iii) because of the permanent changes in an IT-infrastructure of a given organization,
and of the changes arising in its environment as well, such a domain-ontology
has to be constructed as an adaptive ontology;
(iv) to provide a possibility for knowledge gathering and their reusing in ITSMS,
some e-Learning models and technologies can be applied.
There are already the approaches elaborated to combine a CBR-method with on-
tologies [16, 17], which allow to provide more efficiently a case-representation, to
enhance case-similarity assessment, and to perform case-adaptation process for a new
solution. From the other hand, an ontology-centered design for ITSM-services, and
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 57
especially, for Incident Management (IM), is also discussed in some recent publica-
tions in this domain. In particular, the proposed in [21] Onto-IM ontology is built
according to ISO/IEC20000 for ITIL/ITSM [2], it includes such concepts as Incident
Management, Incident Record, Incident Entity, etc. specified using OWL (Ontology
Web Language), and the small example of its notation is shown in Fig.3.
Fig. 3. The excerpt of Onto-IM ontology elaborated in [21]
These results provide a solution for the tasks (i)-(ii), but in our opinion to cover the
task (iii) in efficient way, with respect to permanent changes in IT-infrastructure of a
target organization, an appropriate ontology has to be constructed as an adaptive facil-
ity [22]. In this way the Onto-TO ontology given in Section 4 should be given as the
following tuple
Onto TO ( adapt ) C , R , P , W ( C ) , W ( R ) (16)
where, additionally to the basic components of any ontology, namely: C – set of con-
cepts, R – a set of relationships among these concepts, and P – a set of axioms (se-
mantic rules), the following ones have to be defined: W(C) is a set of weight coeffi-
cients for concepts of C, and a W(R) is a set of weight coefficients for relationships of
R respectively. Usage of these weight coefficients allows us, e.g., to take into account
an appropriate importance grade in several types of ITSMS-customers (see Fig. 2) to
provide IM - services for them.
In order to get all information resources needed for a completed solution of tasks (i)-
(iv), we propose to apply some e-Learning models and technologies within an
ITSMS, especially, for skills training and experience gathering by ITSMS-staff, des-
ignated in the Onto-IM ontology as Incident Manager, ServiceDeskEmployee, Spe-
cialist [21]. For this purpose an e-Learning ontology (Onto-EL) can be used, e.g., in
[23] the Onto-EL is elaborated to build for learners their personal paths in e-learning
environment, according to the selected curriculum (Incident Management in terms of
Onto-IM), syllabus (Incident Record) and subject (Incident Entity).
Summarizing aforementioned issues concerning the tasks listed in (i)-(iv), the
conceptual mechanism to provide an advanced incidents management (AIM) in
ITSMS can be represented at the high-architectural level as the UML-package dia-
gram shown in Fig.4.
Below the approach to resolve the task (1) from their list given in Section 2 is il-
lustrated using the real case-study within our research and practice activities to apply
ITSMS to manage IT-infrastructure of National Technical University “Kharkov Poly-
technic Institute” (www.kharkov.ua) referred in following as NTU “KhPI”.
58 N. Tkachuk, V. Sokol and K.Glukhovtsova
Fig. 4. The AIM – architectural framework (to compare with the scheme given in [24])
6 University Case-Study: Effective ITSM-Modules Configuring
It is to mention that exactly university- and /or a campus-domains are considered from
many authors as a suitable example of ITSMS-usage (see, e.g., in [25-26]), because
intensive research- and educational activities obviously require a modern and well-
organized IT-environment. That is why we also proved our approach to effectiveness
estimation of alternative configurations of ITSM-modules using the test-case data
collected at the NTU “KhPI”.
6.1 Application domain description: IT-infrastructure of NTU “KhPI”
The NTU “KhPI” is one of the largest technical universities of Ukraine located in the
city of Kharkiv, which is the important industrial and cultural center at the East of the
country. The university has about 22000 students, ca. 3500 of faculty members, and
accordingly there is the advanced IT-infrastructure to support all educational and
research tasks. The main characteristics of IT-operating at the NTU “KhPI” are sum-
marized in Table 3.
Table 3. Some technical data about IT-infrastructure NTU “KhPI”
Parameters Values
PCs in the network configuration 1525
User’s accounts 2700
Buildings 23
Servers 60
Routers 80
Peripheral units 6000
IT-specialists in the central office 11
Incidences per day (registered) 5-7
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 59
In cooperation with the IT-staff at the University IT control office we have
analyzed retrospective data about some typical problem situations occurred, and about
the corresponded incidents, which daily have been resolved within the direct
communication with IT-service customers. In this way the main types of ITSM-
incidents and their initial problem situations were identified, and they are described in
Table 4.
Table 4. Main types of ITSM-incidents and their related problem situations
№ Incident type Cause ( problem situation)
1 No Internet-connection at - router was turned off;
Dept or on local PC - network cable breaked or failure on router hardware;
- incorrect network setup;
- problems with software on local PC
№ Incident type Cause ( problem situation)
2 High-loading of PC processor - computer viruses
with a small number of active - high degree of PC hard driver’s de-fragmentation.
user’s programs
3 Installing problems for new - computer viruses
software - absence of additional (middleware) software needed
for installation.
4 Failure to send email -incorrect setup of local network server (proxy)
-problems with central e-mail server.
5 Troubles in the use of third- -lack of specific configuration,
party software - improper use of system services.
Basing on the analysis results obtained, we can apply the elaborated method to es-
timate alternative ITSMS-module configurations (see Section 3).
6.2 Customizing of the elaborated estimation method: alternative
configurations and criteria definition
According to the Step 1 of the method presented in Section 2.2, the list of alternative
ITSM-module configurations have to be defined, and in our case they are presented:
X 1 = Service Desk subsystem (SDS) and Incident Management Module
X 2 = SDS, Incident Management Module and Configuration Management Mod-
ule
X 3 = SDS, Incident Management Module and Change Management module
X 4 = SDS, Incident Management Module and Problem Management Module
On the next Step 2, according to the formulas (4) - (10), we determine the criteria
for the quantitative evaluation of the proposed alternatives and their performance
indicators, which are shown in Table.5. These criteria and their indicators (metrics)
are taken from [35], and they are recommended to evaluate effectiveness of IT-
infrastructure in any business organization.
60 N. Tkachuk, V. Sokol and K.Glukhovtsova
Table 5. List of values for global and local criteria (fragment)
Global and Semantics performance measurement Insecure Effec- Scope of
local criteria criteria and target values value tive values
value
K1 Effectiveness of incident management
k11 Average time incident resolution →min >30 15 9999min.
min. min.
k12 Percentage of incidents resolved proac- 0% 15% 0-100%
tively →max
Global and Semantics performance measurement Insecure Effec- Scope of
local criteria criteria and target values value tive values
value
k13 Percentage of incidents resolved at the <65% 85% 0-100%
first level of support →max
k14 Percentage of incidents that have been <75% 90% 0-100%
resolver from the first time →max
K2 Effectiveness of problem management
k 21 The ratio of the number of solved prob- <10% 35% 0-100%
lems to total problems (%)→max
…….. ……….
For example, a value of 10 for an alternative X 3 to criteria k14 (see Table 5) means,
that the implementation of Service Desk and Incident Management Module will help
to increase the ratio of incidents, which are resolved successfully, to its effective
value of 90%, etc. The obtained in this way results are given in Table 6.
Table 6. Estimated values for the alternatives with respect to the defined criteria (fragment)
K1 : Effective incident management → opt
k11 (opt =20м) k12 (15%) k13 (85%) k14 (90%)
X1 5 5 5 6
X2 6 7 6 6
X3 5 5 5 6
X4 7 6 8 7
… ……………
… … …
In order to implement the elaborated method with customized data introduced
above, the special software tool was developed.
6.3 Results of estimation and their analysis
To continue the usage of our method presented in Section 2.2 (Step 3 and Step 4 re-
spectively) using the pair-wise comparison the weight coefficients of relative impor-
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 61
tance (WCRI designated as w( ki , j ) ) for the local criteria regarding their global ones
were determined:
The WCRI values of the local criteria for the global criterion K1 : w(k11 ) =
0,239458, w(k12 ) = 0,239458, w(k13 ) = 0,432749, w(k14 ) = 0,088335
The WCRI values of the local criteria for the global criterion K 2 : w(k 21 ) =
0,68334, w(k 22 ) = 0,19981, w(k 23 ) = 0,11685
The WCRI values of the local criteria for the global criterion K 3 : w(k31 ) =
0,332516, w(k32 ) = 0,527836, w(k33 ) = 0,139648
The summarized WCRI values for the global criterion K i : K1 = 0,527836, K 2 =
0,332516, K 3 = 0,139648
And finally, according to Step 5 of this method (see Section 3.2), and using the
multi-criteria ranking formulas (11) - (12), we obtain the following ultimate results of
the effectiveness assessment for the considered alternatives (see Table 5), namely
X 1 0.537, X 2 0.671, X 3 0.578, X 4 0.727 (17)
To confirm the reliability of the results given in (17), the comparative analysis with
some "best practices" in ITSMS implementation was carried out, using the data of
IDC-company [28]. In particular, IDC has reviewed approx. 600 organizations
worldwide, which used ITSM for over a year, and in this study especially the prioriti-
zation issues of different ITSM-modules implementation were analyzed. In Fig. 5 the
result of the performed comparison is shown.
Fig. 5. Graphical representation of the obtained results
62 N. Tkachuk, V. Sokol and K.Glukhovtsova
As we can see, to provide Change Management and Configuration Management is
necessary to have within an IT-infrastructure database (DB) of IT-configurations, and
DB of problem situations as well, these facilities are rather costly for the University,
and therefore the implementation of these modules is not a priority task. The most
effective ITSM-modules configuration for NTU "KhPI" includes an Incident Man-
agement module and a Service Desk subsystem.
7 Conclusions and Future Work
In this paper we have presented the intelligent approach to increase efficiency of
ITSMS, which has to resolve 3 interconnected tasks for its effective usage in a target
organization: 1) providing an effective configuration of ITSM-modules according to
its specific features and needs; 2) elaboration an integration framework for a given
ITSMS with existing EA; 3) advanced incidents management in ITSMS. To solve
these tasks in a comprehensive way the interdisciplinary framework is elaborated,
which includes: the expert method for multi-criteria ranking of alternative ITSM-
modules configurations, the ontological specifications for ITSMS-EA integration, and
the approach to enhanced incident management based on the combination of adaptive
ontologies and CBR-methodology. To implement the first part of this approach the
appropriate software tool was elaborated, and its applicability was tested successfully
within the case-study at the NTU “Kharkov Polytechnic Institute”.
In future we are going to implement and to test the appropriate software solutions
for other tasks in the proposed framework, using such technologies as OWL, BPMN,
XML /XLST, and Web-services.
References
1. Office of Government Commerce: ITIL Library. London (2003)
2. International Organization for Standardization. ISO/IEC 20000-1,2: Information Technol-
ogy-Service Management, Part 1, 2. Geneva, Switzerland: ISO/IEC (2005)
3. Braun, C., Winter, R.: Integration of IT Service Management into Enterprise Architecture.
In: Proceeding of SAC’07, Seoul, Korea (2007)
4. ITSM Frameworks and Processes and their Relationship to EA Frameworks. In: A White
Paper by: R. Radhakrishnan, IBM Global Technology Services (2008)
5. Valiente, M.-C., Vicente-Chicote, C., Rodriguez, D.: An Ontology-based and Model-
driven Approach for Designing IT Service Management Systems. In: Int. Journal of Ser-
vice Science, Management, Eng. and Techn., 2(2), pp. 65--81 (2011)
6. Valiente, M.-C., Barriocanal-Garcia E., Sicilia, M.-A.: Applying an Ontology Approach
to IT Service Management for Business-IT Integration. In: Knowledge-Based Systems,
vol., 28, pp. 76--87 (2012)
7. Official Web-site of the Protocol, Ltd. company, http://www.protocolsoftware.com/hp-
openview.php
8. Official Web-site of the BMC Software company, http://www.bmc.com/products/ rem-
edy-itsm/solutions-capabilities/it-service-management-suite.html
9. Official Web-site of the OTRS Group company, http://www.otrs.com
An Intelligent Approach to Increase Efficiency of IT-Service Management Systems 63
10. Official Web-site of the SourceForge code repository, http://sourceforge.net
11. Tkachuk M. V., Sokol V.Y.: Some Problems on IT-infrastructure Management in Enter-
prises: State-of the-Art and Development Perspective. East-European Journal on Advanced
Technologies, 48 (6/2), 68–72 (2010) (in Russian)
12. Official Web-site of the Cleverics company, http://www.cleverics.ru/en
13. Saaty, T. L.: Fundamentals of the Analytic Hierarchy Process. RWS (2000).
14. Jabrailova Z. Q.: A Method of Multi-Criteria Ranging for Personnel Management Problem
Solution. In: Artificial Intelligent, 56 (4), pp.130–137 (2009) (in Russian)
15. Valiente, M.-C., Barriocanal-Garcia E., Sicilia, M.-A.: Applying Ontology-Based Models
for Supporting Integrated Software Development and IT Service Management Processes.
In: IEEE Transactions on Systems, Man, and Cybernetics, Part C: Applications and Re-
views, 42(1), 61–74 (2012)
16. Lopez-Fernandez, H., Fdez-Riverola, L., Reboiro-Jato, M.: Using CBR as Design Meth-
odology for Developing Adaptable Decision Support Systems. University of Vigo, Spain,
pp. 123–145 (2011)
17. Prentzas, J., Hatzilygeroudis, I.: Combinations of Case-Based Reasoning with Other Intel-
ligent Methods. Int. J. of Hybrid Intelligent Systems, 55–58 (2009)
18. Rodríguez-García, D., Barriocanal, E., Alonso, S., Nuzzi, C.: Defining Software Process
Model Constraints with Rules Using OWL and SWRL. J. of Soft, Eng., Knowl. Eng.,
20(4), 533–548 (2010)
19. Prieto, A. E., Lozano-Tello, A.: Use of Ontologies as Representation Support of Work-
flows. J. Network and Systems Management, 17(3), 309–325 (2009)
20. Kang, D., Lee, J., Choi, S., Kim, K.: An Ontology-based Enterprise Architecture. J. Expert
Systems with Applications, 37(2), 1456–1464 (2010)
21. Pansa, I., Reichle, M., Leist, C., Abeck, S.: A Domain Ontology for Designing Manage-
ment Services. In: Proc. 3d Int. Conf. on Advanced Service Computing, pp. 11--18 (2011)
22. Litvin V.: Multi-Agent Decision Support Systems Based on Precedents that Use of Adap-
tive Ontology. Artificial Intelligent, 54(2) 24–33 (2009) (in Ukrainian)
23. Chung, Н.-S., Kim, J.-M.: Learning Ontology Design for Supporting Adaptive Learning in
e-Learning Environment. In: IPCSIT-2012, vol. 27, Singapore, pp.148–152 (2012)
24. Suh, H., Lee, J.: Ontology-Based Case-Based Reasoning for Engineering Design. Design
Research Group Manufacturing Engineering Lab (2008)
25. Boursas, L.: Efficient Technical and Organizational Measures for Privacy-Aware Campus
Identity Management and Service Integration. In: Proc. EUNIS’06, Tartu, Estonia (2006)
26. Knittl, S., Hommel, W.: SERVUS@TUM: User-Centric Service Support and Privacy
Management. In: J.-F. Desnos., Y. Epelboin (eds.) EUNIS’07, Grenoble, France (2007)
27. Brooks, P.: Metrics for IT-service Management. Van Haren Publishing (2006)
28. Official Web-site of the International Data Corporation (IDC), http://www.idc.com
Refining an Ontology by Learning Stakeholder Votes
from their Texts
Olga Tatarintseva1,2 and Vadim Ermolayev1
1
Department of IT, Zaporozhye National University,
66 Zhukovskogo st., 69063, Zaporozhye, Ukraine
tatarintseva@znu.edu.ua, vadim@ermolayev.com
2
Satelliz, 158 Lenina st., P.O. Box 317, 69057, Zaporozhye, Ukraine
Abstract. This paper reports on our experiments evaluating the improvement of
OntoElect approach to ontology refinement in the case study with the ICTERI
Scope Ontology. OntoElect is based on collecting and assessing the commit-
ment of domain knowledge stakeholders for ontological refinement offerings.
We report the improvement with respect to the previous results. Our first ex-
periment evaluates the change in the quality of ontology due to the involvement
of domain knowledge stakeholders in semantic annotation of their papers, com-
pared to the previous study in which the annotations were done by knowledge
engineers. Our second experiment checks if the result became better after the
introduction of the automated term extraction from the full texts of ICTERI pa-
pers. Extracted terms are compared to the manual annotations. The results of
the experiments verify the proposed ontology changes and are further used for
the ICTERI Scope ontology refinement.
Keywords. ICTERI Scope ontology, ontology engineering, domain knowledge
stakeholder, term mining, evaluation, refinement
Key terms. KnowledgeEngineeringMethodology, SubjectExpert, SubjectDo-
main, Metric, Ontology
1 Introduction
Maintaining an ontology in its lifecycle that fits all the requirements of the subject
domain stakeholders is a complicated task in ontology engineering which does not
have a complete solution so far. The problem is to a large extent in devising a meth-
odology for ontology refinement that enables a complete and timely account for those
requirements and maps them to the updated revision of the ontology. One complica-
tion is that the stakeholders who own the requirements need to be committed to pro-
vide their inputs for ontology refinement. Furthermore, those inputs need to be meas-
Refining an Ontology by Learning Stakeholder Votes from their Texts 65
ured and applied correspondingly to the utility of their contribution and in a harmo-
nized way to ensure the consistency and validity of result.
This paper reports on the improvement of our OntoElect approach for iterative on-
tology refinement [1, 2]. The approach has been proposed in [1] using the allusion of
elections in which different “ontology offerings” compete for the commitment of the
pool of the relevant domain knowledge stakeholders being the “electorate”. OntoElect
has been basically validated in an experiment reported in [2] where the approach was
detailed by offering voting metrics for the ICTERI ontology built and refined itera-
tively based on semantic annotation of the pool of papers of ICTERI 2011 conference.
The results of our previous experiment suggested several important technical as-
pects [2] for improving OntoElect methodology as a whole and the accuracy of our
measurements in particular. Some of those aspects have been addressed in the work
reported in this paper.
Firstly, our previous experiment was based substantially on the manual annotation
of papers. A knowledge engineer assigned key terms or suggested missing terms
based on her personal interpretation of the abstract of a paper. By that we mimicked
voting by paper authors while keeping them free of extra annotation effort. The low-
lights of this approach to annotation were that:
Domain knowledge stakeholders (paper authors) were in fact not involved in the
workflow and therefore not motivated to be committed to the resulting ontology re-
finement
The quality of semantic annotations we obtained has been perceived as fairly low
because (a) done by a knowledge engineer who is not a subject expert with respect
to the annotated paper; (b) the source for this work was just an abstract, not a pa-
per, and its meaning has been interpreted by a knowledge engineer.
To overcome those shortcomings we first decided to involve the authors more ac-
tively by requesting that they themselves semantically annotate their submissions to
ICTERI 20121. It has also been considered as promising to refine the approach by
automated extraction of terms from the papers authored by our subject domain
knowledge stakeholders. Here we present the results of our experiments which
checked how these two refinements helped improving the quality and adequacy of the
ICTERI Scope ontology.
Further, the document corpus used in the previous experiment was fairly small in
size for assuring reliable judgements about the opinion of the stakeholder community.
For improving on that we continued the collection of ICTERI papers which has been
extended by all papers of ICTERI 2012.
We first repeat the previous experiment [2] based however on the document corpus
of ICTERI 2012 papers semantically annotated by their authors. We then focus on
answering the question about annotation quality by: (i) performing automated term
extraction from the full texts of our complete document corpus (ICTERI 2011 and
2012); and (ii) comparing the results of automated term extraction to the outputs of
manual semantic annotation.
1
See http://isrg.kit.znu.edu.ua/icteriwiki/index.php/ICTERI-Terms
66 O. Tatarintseva and V. Ermolayev
The remainder of the paper is structured as follows. Section 2 briefly reviews the
related work in relevant fields. Section 3 outlines the OntoElect approach to ontology
refinement and presents the case study dealing with ICTERI Scope Ontology as well
as the document corpus at our disposal. Section 4 sets up our experiments by describ-
ing the workflow, evaluation metrics, and used tools. Section 5 presents and discusses
the results of our experiments. The paper is further concluded and our plans for the
future work are outlined.
2 Related Work
One of the possible ways to check if a conceptualization of a domain is correct and
complete is to evaluate the model against the interpretation of the meaning of the
representative set of relevant documents. The document corpus will be relevant and
representative if it covers the majority of the views by the domain knowledge stake-
holders. Their interpretations may be collected and further analysed for refining the
ontology using different techniques which may be sought in several areas of research
and development. In this section we briefly outline the related work in the relevant
fields of research and refer to our previous publication [1] for a more in-depth and
detailed coverage.
One of the popular relevant research areas studying how interpretations are col-
lected is collaborative or social tagging and annotation. A good survey of the field is
[3] where the use of tags for different purposes and associated shortcomings are ana-
lysed. Semantic annotation and tagging approaches further refine social tagging tech-
niques by offering the collections of terms that are taken from taxonomies, folksono-
mies, or thesauri [4]. Hybrid approaches for collaborative tagging and annotation
aiming at the enrichment of seed knowledge representations by a user community are
reported for example in [5].
One of the promising approaches focused, besides collecting interpretations or sub-
jective conceptualizations, on motivating more people to take part in developing or
refining ontologies is offering a game with a purpose to intended users. Following this
approach, ontology development or refinement can be implicitly embedded in a game
software. There ontology elements are created, updated, and validated implicitly in
the background [6]. Gaming approach has also been tried for evaluating how well
ontological specifications fit to the interpretations of random users (FACTory Game
by Cycorp, http://game.cyc.com/). Several game scenarios have been developed [5]
for ontology building and refinement, ontology matching, annotating content using
lightweight ontologies. Those are similar to our OntoElect approach. Both approaches
offer possibilities to identify whenever users start to agree on and share commitment
to certain ontological items.
Social and gaming approaches that involve the direct participation of human stake-
holders are complemented by the plethora of research results in automated knowledge
extraction or ontology learning. This strand of research involves the stakeholders
indirectly – through making use of their professional outputs, like authored texts. A
comprehensive survey of the techniques used to learn ontologies from texts is [7]. In
Refining an Ontology by Learning Stakeholder Votes from their Texts 67
the second experiment we present in this paper only term extraction using the Ter-
Mine tool [8] has been performed.
Yet one more important aspect in developing or refining an ontology is the re-use
of the other ontologies or their most relevant parts to the developed ontology. In this
context the ontology meaning summarization approach [#] makes good sense for
helping an ontology engineer choose the most relevant and valuable parts for re-use.
The approach is based on detecting the “key concepts” of an ontology under analysis
which best characterize its meaning. The key concepts are determined using a combi-
nation of criteria from lexical statistics, taxonomy graph analysis, and popularity
based on a number of hits. Especially in using the popularity and coverage metrics,
this approach coincides well with our approach (OntoElect). OntoElect is however
used not for summarizing but refining an ontology based, among other things, on
assessing the coverage and popularity of the Key Terms. Besides that the mechanisms
of obtaining the measures are different. From the other hand, OntoElect does not yet
consider ontology re-use as one of important mechanisms for refinement. Hence,
combining some features of [9] in OntoElect may be enriching.
3 ICTERI Case Study
The idea of OntoElect approach [1] was inspired by public election campaigns. Just as
the leader in a public election campaign gets the major part of the electorate’s com-
mitment to win, the extent of the domain knowledge stakeholders' commitment hints
about the quality and completeness of the ontology. Following this allusion, the votes
of the domain knowledge stakeholders for alternative ontology offerings are collected
and used as the measure of their commitment. The ontology offering that collects the
biggest share of votes could therefore be considered as the best and most complete.
In our case study the OntoElect approach is applied for refining the ICTERI Scope
Ontology in the iterative ontology engineering experiment. Our domain knowledge
stakeholders are the authors of ICTERI papers. Ontology offerings in the reported
work are the strucrtural contexts2 in the five thematic areas of the ICTERI scope of-
fered to the authors for choosing the appropriate key terms to annotate their papers.
As this data had to be selected we decided to simulate the opinions of the electorate
by annotating the papers of ICTERI 2011 manually. For this we extracted the terms
which were specified as the list of ICTERI Key Terms if it was possible. In some
cases we had to add Missing Concepts (also called Missing Key Terms) for the pa-
pers, if such terms did not exist in the list.
So, we received three semantic annotation types:
KeyWord – for the key words, which were selected by the authors
KeyTerm – for the terms which were selected manually and were found in the list
of the ICTERI terms
2
A structural context, as suggested e.g. in [10], is composed of a central concept with all his
domain and object properties and the concepts connected to the central concept by these ob-
ject properties.
68 O. Tatarintseva and V. Ermolayev
MissingConcept - for the terms, which did not exist in the list of the ICTERI terms,
but were covered during annotation
One use of the particular term was considered to be one vote for the selected term.
The votes were normalized as frequencies of use. Such information allowed us to
measure the popularity of each semantic context, circumscribe the most frequently
demanded part of the ontology and make suggestion about the completeness of the
ontological offerings.
For the papers of ICTERI 2012 we requested that the authors annotate their papers
not only using the freely chosen key words, but also using the terms found in ICTERI
scope ontology. As the result the corpus for the futher analysis was increased. The
data provided by the authors can be accepted as more authentic than that which was
obtained ourselves, as they are the real domain experts for the field they study. The
analysis of the received data is presented in Section 5.
But even when we use the information presented by the authors, and the results of
our manual annotation we can't guarantee that this information is accurate enough for
applying it to the ontology refinement process. To obtain the experiment we needed to
have results received in different ways because we wanted to achieve the impartial
assessment of OntoElect approach. Before applying the results in ontology refining
process we decided to check them with freely available tool for text mining.
For our experiment we chose one of the services provided by the National Centre
for Text Mining (NaCTeM). As reported in the official website of NaCTeM3 it is the
first publicly-funded text mining centre in the world. It provides text mining services
in response to the requirements of the UK academic community. NaCTeM is operated
by the University of Manchester.
4 Experimental Set-up and Tools
To control the results of the experiment we have to understand which main questions
we are going to answer after its realization and how to measure these results.
Our measurable objectives for the experiment have been formulated as follows [2]:
Does the ontology fit to the requirements of the subject experts in the domain?
The fitness of the ontological offering will be measured as a ratio of the average
frequency of use of the available Key Terms (positive votes) to the similar for the
missing Key Terms (negative votes). Special attention will be paid to the freely cho-
sen key words that are identical to the available Key Terms. Those will be considered
as extra positive votes for the semantic context of the Key Term.
Is there a particular part in the ontology that is the most important for the stake-
holders?
The importance of an ontology fragment comprising particular concepts will be
measured as frequency of use of these concepts (positive votes). Fragments of differ-
ent importance will also be presented as percentiles.
3
Official web site of National Centre for Text Mining http://www.nactem.ac.uk/
Refining an Ontology by Learning Stakeholder Votes from their Texts 69
Is there a part in the ontology that could be dropped as the stakeholders do not
really require it?
Similarly to importance, these ontology fragments will be outlined using low fre-
quency of use percentiles.
What would be a most valuable addition to the ontology that will substantially
improve stakeholders’ commitment to it?
The papers have been annotated using missing Key Terms and freely chosen key-
words. Those missing Key Terms that are frequently used will form the core of this
effective extension. If some of the keywords are also used frequently by the authors
they may become good candidates for the inclusion in the effective extension as well.
Special attention will be paid to the freely chosen key words that are identical to the
missing Key Terms. Those will reinforce the votes on the addition to the ontology.
The flow of activities has been organized in three consecutive phases as presented
in Fig. 1. The description of each phase in details is presented in [2].
Conference - Annotate using
Management System ICTERI MediaWiki ICTERI Scope ICTERI Semantic MediaWiki
(EasyChair) Ontology concepts
- Parse - Collect missing ICTERI Key Terms
- Add Wiki mark-up concepts
- Import to ICTERI - Collect free key
Wiki words Semantic annotations (votes)
Missing Key Terms
PHASE 1 PHASE 2
- 2011: ontology
engineer
Submissions Data ICTERI paper
Corpus pages
Legend:
- Activity is performed using the developed - Analyse votes
tool using SMW queries
- Activity is performed manually ICTERI papers
PHASE 3
- ontology Keywords
engineer
Fig. 1. The workflow for processing ICTERI papers and collecting stakeholders’ votes repeated
according to [2]
At phase 1 we have extracted the semi-structured information about the papers ac-
cepted for ICTERI 2012 and transformed these into the collection of paper articles in
the ICTERI Wiki. At Phase 2 we extracted the freely chosen KeyWords and the
KeyTerms from the ICTERI Scope ontology assigned by the authors and added these
to the semantic annotations of the papers. In several cases we detected considerable
meaning gaps between the extracted key words and Key Terms when annotated the
papers manually. Therefore, we opted to add the missing Key Terms to the corre-
sponding semantic annotations. As a result of this Phase the semantic relationships
between the pages of Category:Paper and the pages of Category:Concept
have been specified as semantic properties. These semantic properties allowed us to
receive all the measurements planned for the evaluation experiment. These measure-
ments have been done using different Semantic MediaWiki queries at Phase 3.
70 O. Tatarintseva and V. Ermolayev
Compared to the previous year experiment [2], we automated the extraction of the
frequency of use statistics which made the process less error prone and faster. For that
the SMWAskAPI4 extension of the Semantic MediaWiki has been used. This exten-
sion supports semantic queries of #ask and enables the use of the corresponding API
for executing Semantic MediaWiki ask queries.
Each page of the ICTERI Wiki uses semantic tagging. An example of the Semantic
properties specified for pages in Category:Paper is given in Fig. 2.
For our analysis we used the pages from Category:Paper and Cate-
gory:Workshop with the property hasPublicationYear equal to 2012, namely the
values of the semantic properties hasKeyWord, hasKeyTem, and MissingConcept.
Fig. 2. Semantic properties for the ICTERI Wiki page in the Category:Paper
The scripts for analyzing these values were coded in Python. Some steps were also
implemented using shell scripting. As outputs we have received:
The list of KeyWords, KeyTerms, and MissingConcepts for each article, if they
were defined
The overall number of the papers according to the values of the properties hasPub-
licationYear, and the selected Category
The number occurrences of each KeyWord, KeyTerm and MissingConcept.
The analysis and discussion of the results is given in Section 5.
To perform our second experiment we applied the Term Management System
named TerMine, which identifies key phrases in text. It uses C-value [8], a domain-
independent method for automatic term recognition (ATR) which combines linguistic
and statistical analyses with the emphasis on the statistical part. The linguistic analy-
sis enumerates all candidate terms in a given text by applying part-of-speech tagging,
extracting word sequences based on adjectives/nouns, and stop-list. The statistical
4
See the description of the SMWAskAPI on http://www.mediawiki.org/wiki/Extension:
SMWAskAPI
Refining an Ontology by Learning Stakeholder Votes from their Texts 71
analysis assigns a candidate term to a termhood by using the following four character-
istics:
The occurrence frequency of the candidate term
The frequency of the candidate term as part of other longer candidate terms
The number of these longer candidate terms
The length of the candidate term
The data corpus for this term extraction and analysis was the merge of the pools of
ICTERI 20115 and ICTERI 20126 papers published in the respective proceedings, and
consisted of 63 papers. The papers from both proceedings volumes have been merged
in a single file and uploaded for processing by TerMine. The workflow for the second
experiment is pictured in Fig. 3.
ICTERI 2011 File
proceedings upload Term
extraction
using Statistical results
ICTERI TerMine for further analysis
proceedings
ICTERI 2012
proceedings
Fig. 3. The workflow for conducting the second experiment for term mining and analysis
The data processed in the pipeline is illustrated by the example of a single
paper [1] in Fig. 4.
The results of term mining were provided in several forms. All the terms defined in
the text were highlighted by colour markings (upper part of Fig. 4a). The information
about the overall number of the terms mined from the text was also given (433 terms
listed – in the bottom of Fig. 4a). The terms were also presented in the table view,
each preceded with the assigned rank number and followed by the statistical score
measure (lower part of Fig. 4a). The rank of a term means the position of each term in
the table sorted by the score; the rank values of the terms with the same score are
equal. The scores were computed automatically using the Term Recognition tech-
nique [8] which uses the information about the frequencies of term occurrence. This
approach is essentially a shallow bag of terms extraction technique – therefore the
output needs to be post-processed as described using our single paper data example.
For this example the number of extracted terms was 433 which is obviously too many.
To compare, the authors were advised to assign 3-5 KeyTerms to their papers which
best describe its meaning. Among those extracted terms that we needed to sort out
were also names, affiliations, cities, etc, which had no semantic relationship to the
5
http://ceur-ws.org/Vol-716/
6
http://ceur-ws.org/Vol-848/
72 O. Tatarintseva and V. Ermolayev
meaning of the paper. Also, it has been assumed that the terms with a low number of
occurrences in text have a negligent semantic contribution and may also be filtered
out – so only the higher ranked part of the term list may be considered.
Rank Term Score TimesInText
1 ontology engineering 25,85 27
2 knowledge representation 14,90 16
Term list
3 subject expert 11 12
purifying
4 intended user 10 12
5 ontology element 7 7
6 psi suite 6 6
6 stakeholder commitment 6 6
6 semantic technology 6 7
9 active involvement 5 5
9 semantic web 5 5
9 ontological context 5 5
(b) 11 high-ranked terms listed Comparative analysis
of achieved results
433 terms listed
(a) Statistical results presented by TerMine (c) Selected KeyWords and KeyTerms by the authors
Fig. 4. An example of the data processed in the term mining experiment
While post-processing the list of mined terms we decided to leave only the terms,
which were used more than 5 times, and had score more than 5 points. Applying this
threshold returned 11 of 433 terms for the example outlined in Fig. 4., which consti-
tutes only 2.54 per cent of the overall number of the mined terms. The manual check
of the example however indicates that these 11 high ranked terms indeed contribute
most significantly to describing the semantics of the corresponding paper (Fig. 4b).
The right part of Fig. 4 allows to compare the result of term extraction (Fig. 4b)
with the output of manual semantic annotation (Fig. 4c) for the selected example pa-
per. A mechanical comparison reveals substantial difference, which however is not
that big after manual mapping of the extracted terms to the concepts of the ICTERI
Scope ontology. In fact there is a subset of extracted terms that could be directly
mapped into the Key Terms of the ontology: subject expert; ontology engineering (as
a methodology). Another group is relevant to the assigned KeyWords: ontology,
stakeholder commitment, ontology engineering. Some are synonymic in the context
of this paper: stakeholder and intended user. Some represent the meaning which is too
fine-grained for a semantic annotation: ontology element. And, which is most impor-
tant, some are the new valid candidates for the inclusion into the ICTERI Scope on-
tology: knowledge representation, semantic technology, semantic web.
5 Results and Discussion
In this section we present the results of the experiment. The set up of all its stages is
described in Section 4. The discussion of the experiment results is structured along
the measurable items.
Refining an Ontology by Learning Stakeholder Votes from their Texts 73
The frequency of use diagrams (Fig. 5 and 6) are built in regard to the total amount
of the papers and the number of occurrences of a particular term.
Similar work was reported in our previous publication [2]. In it we described the
mechanism of ranging. The actual experiment is based on the previous results. But
they changed as the document corpus we are working with has increased and the data
to work with has changed. As we consider OntoElect approach in the case study of
iterative refinement of the ICTERI Scope Ontology such changes are greatly impor-
tant. The diagrams which show these changes are shown below:
For the KeyWords which were selected by the authors manually (only those, which
were chosen by at least two authors, Fig. 5)
For the KeyTerms, which were selected from the ICTERI ontology terms (Fig. 6)
Algebraic programming
Competence
Competence Formation Process
Computational experiment
Computer Science
Design
Development
Distance learning
Distance learning system
Dynamic Systems
Educational Process
E-learning
Evaluation
Feedback
Grid
Informatics
UsedKeyWord
Information
Information and communication
Information System
Information technology
Informatization
Insertion Modeling
Internet
IT Specialist
Labor Market
Object-oriented Approach
Ontology
Ontology Engineering
Partial Predicates
Reachability
Software
Software Design
Software development
Sorting algorithms
Teaching Process
Verification
Virtual Class
0 1 2 3 4 5 6 7 8 9 10
FrequencyOfUse, %
ICTERI 2011 KeyWord ICTERI 2012 KeyWord
Fig. 5. The frequency of use of the freely chosen KeyWords
We did not provide a frequency of use diagram comparing the Missing Key Terms
because the difference in the results of 2011 and 2012 is tiny and could be neglected.
We provided the comparison analysis for KeyWords and Missing Key Terms lists
(Fig. 7). To compute the range of use of each term we divided each frequency of use
index by the frequency value of the most popular term, which range of use was taken
as 100 per cent. These terms are not the part of the ontology, but are the most possible
candidates.
74 O. Tatarintseva and V. Ermolayev
Academia
Agent
Approach
BioInspiredApproach
BusinessIntelligence
Capability
Characteristic
Collaboration
Competence
CompetenceFormationProcess
Computation
Computing
Cooperation
Data
DecisionSupport
Dev elopment
Didactics
Env ironment
Ex perience
FormalMethod
Grid
GridInfrastructure
HighPerformanceComputing
ICTComponent
ICTEnv ironment
ICTTool
ImmaterialArtifact
Industry
InformationCommunicationTechnology
InformationTechnology
Infrastructure
Integration
Intelligence
Know ledgeEngineeringMethodology
Know ledgeEngineeringProcess
Know ledgeEv olution
Know ledgeManagementMethodology
UsedKeyTerm
Know ledgeManagementProcess
Know ledgeRepresentation
Know ledgeTechnology
Know ledgeTransfer
LaborMarket
LinkedData
MachineIntelligence
Management
MathematicalModel
MathematicalModeling
Method
Methodology
Metric
Model
ModelBasedSoftw areDev elopmentMethodology
MultiAgentSy stem
Need
Nominativ eData
PSI-ULO:Object
PSI-ULO:Process
PSI-ULO:ProcessPattern
Qualification
Quality AssuranceProcess
Qualty AssuranceMethodology
Reasoning
Requirement
Research
SemanticWebServ ice
Socially InspiredApproach
Softw areComponent
Softw areEngineeringProcess
Softw areSy stem
SpecificationProcess
StandardizationProcess
SubjectDomain
SubjectEx pert
TeachingMethodology
TeachingPattern
TeachingProcess
Technology
Tool
VerificationProcess
WebServ ice
0 4 8 12 16 20 24 28
FrequencyOfUse, %
KeyTerm 2011 KeyTerm 2012
Fig. 6. The frequency of use of available KeyTerms
Communication
ComputationalEx perime
ConceptualModeling
Constraint
DataSy nchronization
DecisionMaking
DistanceLearningSy stem
Distribution
EducationalTool
ELearningSy stem
Ev idenceModeling
GridComputing
Term
InformationSy stem
InsertionModeling
IntensionalizedData
Logic
Negotiation
Ontology
Ontology Engineering
Quality ManagementSy st
Serv iceNegotiation
Softw are
Softw areDev elopment
TaskAssignment
Verification
0,00 20,00 40,00 60,00 80,00 100,00
Range, %
KeyWord MissingConcept
Fig. 7. The range of use for Missing Key Terms and KeyWords
Refining an Ontology by Learning Stakeholder Votes from their Texts 75
The results which we received at the previous step can be merged and we can re-
ceive the new version of the potential ontology offering. But before doing this we will
look into the results of the text mining experiment.
Using TerMine tool for automation of the knowledge mining process we received
some interesting results. The overall number of the found terms was 8487. All of the
selected terms were graduated and received the position in the rank table. Studying
the results it is obvious that the number of the terms selected by the data mining tool
is too big.
It was decided to leave in the rank table only those terms which have the score of
10 and more. The popularity of the terms which were picked up is evident. The total
number of such terms is 157. It is easy to count that it makes up less than 2% of all
the terms proposed by the tool. After refining the list and deletion of the superfluous
information only 140 terms left.
To compare the frequencies of use provided by the TerMine and our own calcula-
tions we decided to use percentage method (similar to that used for building the dia-
gram in Fig. 7). We took the maximal value for each group of concepts as 100 per
cent and divided it by the frequency of use value of a particular term. As a result each
term got the value, called the range, which could be compared with the ranges of the
other terms. We analyzed the three groups of mined term matches to the: (i) Key-
Words; (ii) MissingConcepts; and (iii) KeyTerms. As the number of the Missing Con-
cepts is not too big we decided to combine them with the Key Words in the diagram
(Fig. 8).
AbstractQuantumAutomaton
AlgebraicExpression
ComputerScience
ConceptualModeling
ConstraintProgramming
DeadlockAnalysis
DistanceLearning
DistanceLearningSystem
EducationalProcess
EducationalProgrammingLangua
ElectronicDocumentManagement
FinalDemand
FormalVerification
GrossOutput
InformationResource
InformationSystem
Term
InsertionModeling
LaborMarket
LogicalSystem
MultiagentSystem
OntologyEngineering
ProgrammingLanguage
ProofAssistant
QualityManagementSystem
R&DManagement
RussianMusic
SemanticWeb
ServiceNegotiation
SocialNetwork
SoftwareDevelopment
SoftwareEngineering
VariabilityManagement
VirtualLaboratory
0 10 20 30 40 50 60 70 80 90 100
Range, %
TermByTool KeyWord MissingConcept
Fig. 8. The range of use for terms detected by tool, Key Words and Missing Concepts. Range
values were normalized by the frequency of use of the highest scored extracted term (100)
76 O. Tatarintseva and V. Ermolayev
After the automated search of the identical terms in the lists of KeyTerms and
terms mined by tool we discovered that some of them were missed as the search did
not use the rules of common sense and the relations described in the ontology. For
example, according to the ICTERI Scope ontology the term Integration subsumes to
PSI-ULO:Process. Knowing this fact we understand that the term IntegrationProcess
is just the same as the term Integration. But this match is not obvious for the simple
search and will not be detected.
Therefore, to find the matches in the lists of KeyTerms and terms mined by the tool
we decided to undertake a more careful analysis. We scanned the list of the KeyTerms
for matching the ToolTerms manually. Besides for this process we used the whole
pool of 8487 terms mined by the tool. The result is pictured in the diagram (Fig. 9).
Agent
BusinessIntelligence
Characteristic
Collaboration
Competence
CompetenceApproach
Computation
Computing
DecisionSupport
Development
Didactics
Environment
FormalMethod
Grid
HighPerformanceComputing
IctComponent
ICTEnvironment
IctTool
ImmaterialArtifact
Industry
InformationCommunicationTechnology
InformationTechnology
Infrastructure
Integration
Intelligence
KnowledgeEngineeringMethodology
KnowledgeEngineeringProcess
KnowledgeEvolution
KnowledgeManagementProcess
KnowledgeRepresentation
KnowledgeTransfer
LaborMarket
Term
LinkedData
MachineIntelligence
MachineIntelligence
Management
MathematicalModel
Method
Methodology
Metric
Model
ModelBasedSoftwareDevelopmentM
MultiAgentSystem
NominativeData – datum
PSI-ULO:Object
PSI-ULO:Process
QualityAssuranceProcess
Reasoning
Requirement
Research
Semantics
SemanticWebService
SoftwareComponent
SoftwareEngineeringProcess
SoftwareSystem
SpecificationProcess
StandardizationProcess
SubjectDomain
SubjectExpert
TeachingPattern
TeachingProcess
Technology
Tool
VerificationProcess
WebService
0 10 20 30 40 50 60 70 80 90 100
Range, %
TermByTool KeyTerm
Fig. 9. The range of use for the KeyTerms and the terms extracted by the TerMine tool
Refining an Ontology by Learning Stakeholder Votes from their Texts 77
6 Concluding Remarks and Future Work
The paper reported on the experiment evaluating the improvement of OntoElect ap-
proach to ontology engineering in the case study with the ICTERI Scope Ontology. In
particular, the approach has been used to evaluate the validity of the papers’ annota-
tion by drawing knowledge stakeholders to their own papers’ annotation and by
studying their quality using voting for full texts.
The experiment consisted of two stages. The first one was based on the compari-
son analysis of the papers presented during international conferences ICTERI 2011
and ICTERI 2012. The second one was dedicated to performing automated term ex-
traction from the full texts and comparing the results of automated term extraction to
the outputs of manual semantic annotation.
Achieved results stress the important parts of the ontology and those which are
less popular among the authors. The comparison analysis of the first experiment
shows how the situation changed during two years. The KeyWords and MissingCon-
cepts which have high frequency of use values, especially if they are named in both
lists, are the first candidates to become the new part of the ontology.
The second experiment shows which ontological offerings agree with the terms
mined by tool and which numeric characteristics these matches have. The terms ex-
tracted by the tool and their matches with the KeyWords and MissingConcepts, which
have range more than 50 per cent, are also good candidates to be added to the ontol-
ogy. Besides, the degree to which the extracted terms match the KeyWords and
KeyTerms indicate about the adequacy of paper annotation. Overall the overlap be-
tween the meanings of the extracted terms and the KeyTerms measures the range of
so to say the similarity in the meanings of the papers within the corpus and the onto-
logical offerings aimed at covering these meanings. The quantitative results of our
experiments still need to be processed and analyzed more thoroughly before deciding
about the implementation of the changes to the ontology. Besides that several other
aspects still need to be researched in our future work.
Firstly, the document corpus used in the case study, though growing, is still not
very big to allow robustly applying the majority of traditional knowledge extraction
techniques. At the moment it could only be stated that the information we have now is
enough to prove the concept – i.e. the validity of the approach based on the assess-
ment of and account for domain knowledge stakeholder opinions, implicitly reflecting
their needs. After applying the refinements suggested by the stakeholders, the ontol-
ogy still needs to be evaluated and validated using other methods.
Secondly, in this paper we reported about only a partial and shallow way of ex-
tracting knowledge from paper texts. A possible refinement to this preliminary solu-
tion could be sought in using a hybrid iterative knowledge extraction workflow that
incrementally adds ontology elements to the “ontology learning layer cake (c.f.
[11])”.
78 O. Tatarintseva and V. Ermolayev
References
1. Tatarintseva, O., Ermolayev, V., Fensel, A.: Is Your Ontology a Burden or a Gem? – To-
wards Xtreme Ontology Engineering. In: Ermolayev, V. et al. (eds.) Proc. 7-th Int. Conf.
ICTERI 2011, Kherson, Ukraine, May 4-7, 2011, CEUR-WS.org, vol-716, ISSN 1613-
0073, 65–81, online (2011)
2. Tatarintseva, O., Borue, Yu., and Ermolayev, V.: OntoElect Approach for Iterative Ontol-
ogy Refinement: a Case Study with ICTERI Scope Ontology. In: Ermolayev, V. et al.
(eds.) Proc. 8-th Int. Conf. ICTERI 2012, Kherson, Ukraine, June 6-10, 2012, CEUR-
WS.org, vol-848, ISSN 1613-0073, 244, online (2011)
3. Gupta, M., Li, R., Yin, Z., Han, J.: Survey on Social Tagging Techniques. SIGKDD Ex-
plorations 12(1), 58–72 (2010)
4. Uren, V., Cimiano, P., Iria, J., Handschuh, S., Vargas-Vera, M., Motta, E., Ciravegna, F.:
Semantic annotation for knowledge management: Requirements and a survey of the state
of the art. Science. Services and Agents on the World Wide Web 4(1), 14–28 (2006)
5. Hunter, J., Khan, I., Gewrber, A.: HarvANA – Harvesting Community Tags to Enrich Col-
lection Metadata. In: Paepcke A, Borbiha J, Naaman M (eds.) 8th ACM/IEEE-CS Joint
Conference on Digital Libraries, 147–156. ACM New York, New York (2008)
6. Siorpaes, K., Hepp, M.: Games with a Purpose for the Semantic Web. IEEE Intelligent
Systems 23(3), 50--60 (2008)
7. Wong, W., Liu, W., and Bennamoun, M.: Ontology learning from text: A look back and
into the future. ACM Comput. Surv., 44(4), Article 20, 36 pages.
DOI=10.1145/2333112.2333115 (September 2012)
8. Frantzi, K., Ananiadou, S. and Mima, H.: Automatic recognition of multi-word terms. Int.
J. of Digital Libraries 3(2), pp.117-132 (2000)
9. Peroni, S., Motta, E., d'Aquin, M. Identifying Key Concepts in an Ontology, through the
Integration of Cognitive Principles with Statistical and Topological Measures. In: Proc. 3rd
Asian Semantic Web Conference (ASWC 2008), Dec 08-11, 2008, Bangkok, Thailand
(2008)
10. Ermolayev, V., Copylov, A., Keberle, N., Jentzsch, E., Matzke, W.-E.. Using Contexts in
Ontology Structural Change Analysis.. In: Ermolayev, V., Gomez-Perez, J.-M., Haase, P.,
Warren, P, (eds.) CIAO 2010, CEUR-WS, vol. 626 (2010)
11. Wong, W., Liu, W., Bennamoun, M.: Ontology Learning from Text: a Look Back and into
the Future. ACM Comput. Surv., 44(4), Article 20, 36 p.,
http://doi.acm.org/10.1145/2333112.2333115 (2012)
Answering Conjunctive Queries
over a Temporally-Ordered Finite Sequence of ABoxes
Sharing one TBox
Natalya G. Keberle ?
Zaporozhye National University, Dept. of Information Technologies
66, Zhukovskogo str. 69063, Zaporozhye, Ukraine
nkeberle@gmail.com
Abstract. Ontology-based data access (OBDA) assumes that data in a database
are mediated with a conceptual layer, available for clients and hiding data storage
details. Ontologies are good candidates for such a conceptual layer presentation,
whereas databases are good for huge data storage. One of the interesting appli-
cations of OBDA is checking a finite set of constraints defined in some language
against a temporally-ordered sequence of ABoxes sharing one TBox, where each
constraint is considered as a conjunctive query. Presented is one algorithm of con-
junctive query answering for such a language, proved are its termination, sound-
ness and completeness.
Keywords. Ontology-based data access, temporal conjunctive query language,
description logic knowledge base
Key terms. KnowledgeEvolution, KnowledgeManagementProcess, DecisionSup-
port
1 Introduction
Ontology-based data access (OBDA) [1] assumes that data in a database are mediated
with a conceptual layer, available for clients and hiding data storage details. Ontologies
are good candidates for such a conceptual layer presentation, whereas databases are
good for huge data storage.
The benefits from combination of databases and knowledge bases are as follows:
– database management is the mature field of research, there is a lot of commercially
and freely avaliable DBMSs, showing good-to-excellent performance on large datasets.
It is an obvious place to store the assertional part of some knowledge base, i.e. an
ABox;
– a TBox often requires a reasoning support to deduce additional assertions, axioms
and to check the consistency of a knowledge base.
?
The work done during the research visit to Dresden University of Technology, sponsored by
The Ministry of Education and Science, Youth and Sport of Ukraine
80 N. G. Keberle
At the same time, employing such an approach is rather challenging due to signifi-
cant differences between relational database systems and ontology languages, based on
Description Logics, such as OWL. At first, relational databases adopt a closed-world
semantics, i.e. all facts that are not explicitly stated to be true are assumed to be false.
In contrast, OWL is based on an open world semantics which does not requre one to
fix the truth value of every fact and is more similar to an incomplete database. Second,
relational databases are unaware of the intensional part of a knowledge base (called a
TBox).
Research has been done so far in the OBDA field considers only one ABox stored in
a data source, that is an actual set of assertions on individuals and their pairs. However,
real applications show that ABox is changing over time. The examples of such dynamic
systems can be easily found in practice: environmental conditions, air traffic load, com-
puter system load and performance, health control for the people suffering from serious
diseases. Therefore, in some applications of situation awareness [2], there is a need to
store an archive of ABoxes, keeping ABoxes actual at different time points. Temporal
logics are often used as the means to formulate constraints a dynamic system should
obey during its work.
The main results of the paper are:
– for the point-based linear finite time structure elaborated is the language of unions
of temporal conjunctive queries, which allows to evaluate atemporal unions of con-
junctive queries at different time points;
– proposed is an algorithm of answering a union of temporal conjunctive queries,
which harnesses set-theoretic operations on atomic queries answer sets. Proved are
its termination, soundness and completeness;
The paper is organized as follows: in the next section a language of unions of temporal
conjunctive queries is introduced and the definitions of main reasoning tasks available
for such a query language are presented. In the section 3 the algorithm of answering
temporal unions of conjunctive queries is presented and illustrated in examples. The
section 4 is dedicated to the proofs of logical properties of the algorithm. The section 5
discusses the related work in the field of conjunctive queries answering.
2 Conjunctive Queries: Syntax, Semantics
Assume there is a knowledge base K = (A, T ), where T is a set of concept axioms (a
TBox), A is a set of assertional axioms (an ABox). Fix a language of a knowledge base
to ALC [3]. An interpretation of K, named I, is a pair (·I , ∆), where ∆ is a domain of
individuals, obeying unique name assumption (UNA) and ·I is an interpretation func-
tion, which assigns every concept C a set C I ⊆ ∆, every atomic role R a binary relation
RI ⊆ ∆ × ∆, and every individual name a an individual aI ∈ ∆. Assertional axioms
are C(a) - concept assertion and R(a, b) - role assertion.
Query answering is the extension of a well known task of instance checking: given
a knowledge base K and an assertion α. Check whether this assertion is entailed by an
ABox of K.
Answering Conjunctive Queries over a Temporally-Ordered ... 81
2.1 Conjunctive Queries Basics
Let Vars(Q) be a set of all distinguished and non-distinguished variables which appear
in a query Q, let Inds(Q) to denote the set of all individual names which appear in
query Q and Terms(Q) to denote the set of all terms in Q, i.e. Vars(Q) ∪ Inds(Q).
Let us formally define conjunctive queries and Boolean conjunctive queries for a well-
elaborated language ALC [3].
Definition 1 (Conjunctive query, Union of conjunctive queries). Let x, y, c are re-
spectively tuples of distinguished variables (answer variables), of non-distinguished
variables and of individual names, and t, t1 , t2 are terms in Terms(Q). A conjunctive
query (CQ) is an expression of the form
conj(x, c) = ∃y.q1 ∧ . . . ∧ qm ,
where
qi ::= C(t) | r(t1 , t2 )
A Boolean conjunctive query is a CQ without answer variables.
A union of conjunctive queries (UCQ) is a disjunction of conjunctive queries (CQs)
of the form
Q(x) = {x | conj1 (x, c) ∨ . . . ∨ conjn (x, c)}
Example 1. The example of a query asking about all students that attend some courses
and take some exams could be as follows:
Q(x) = {x | ∃y.takeCourse(x, y) ∧ takeExam(x, y)}
This query can be modified to a Boolean query by substitution of x with an individ-
ual name:
Q(x) = {| y.takeCourse(”Eldora”, y) ∧ takeExam(”Eldora”, y)}
We use |Q| to denote the size of Q - the number of symbols required to build the
query. The arity of a query will be the number of answer variables in the query. If
all terms in Q are individual names, we say Q is ground. We write Q(c) for a query
whose answer variables x are substituted by c, Q(x) for a conjunctive query and simply
Q for a Boolean conjunctive query. Sometimes we write x1 , . . . , xn instead of x, and
similarly for y and c.
Given an ALC-knowledge base K = hT , Ai, an interpretation I satisfies a query
Q(x) iff the interpretation function can be extended to the variables in Q(x) in such a
way that I satisfies every term in Q(x). A query Q(x) is true w.r.t. K (written K |= Q)
iff every interpretation that satisfies K also satisfies Q.
Definition 2 (Query answering, query entailment). Given a query Q(x) with a tuple
of answer variables x, and a knowledge base K, a tuple of individuals c with the same
arity of x is an answer for Q in K if I |= Q(c) for every model I in K.
Given a Boolean conjunctive query Q, and a KB K, query entailment is a task to
decide whether K |= Q if I |= Q for every model I of K.
Given a conjunctive query Q(x), a tuple of individuals a, and a KB K, query an-
swering is a task to decide whether a is an answer for Q(x) in K.
82 N. G. Keberle
3 Temporal Conjunctive Queries
Let K = hT , (Ai )0≤i≤n i be a knowledge base with a sequence of ABoxes sharing
one TBox. Let’s describe a query language extending the language of conjunctions of
positive existential formulae built from query atoms. Having in mind linear temporal
logic LT L (see e.g. [4]), this language allows for the following temporal operators: #
(next), #− (previous), U (until), S (since).
Definition 3. Temporal conjunctive query (TCQ) Ψ is an expression
tconj(x, c) = ∃y.q1 ∧ . . . ∧ qm ,
where
qi ::= ϕ | Ψ
ϕ ::= C(t) | r(t1 , t2 )
Ψ ::= ϕ | Ψ1 ∧ Ψ2 | #− Ψ | #Ψ | Ψ1 S Ψ2 | Ψ1 U Ψ2
and C is a concept description, r is a role name, t, t1 , t2 are terms in Terms(Q).
Derived temporal modalities like 3− (sometimes in the past), 2− (always in the
past), 3, 2 can be defined in a usual way (see, e.g. [4]).
Example 2. A query asking about students who had defended their thesis some time
ago and had been ex-matriculated since then is expressed as follows:
Q(x) = {x | ∃y.3− Student(x) ∧ exM atriculated(x) S hasDef ended(x, y)}
The semantics of the TCQ is defined as follows.
Definition 4. A total function π : Terms(Ψ ) → ∆ is a binding for a query Ψ in an
interpretation I, if π(a) = a for all individuals a ∈ dom(π), and the validity I, π |= Φ
for atemporal CQ ϕ is defined inductively:
I, π |= C(t) iff I |= C(π(t))
I, π |= r(t1 , t2 ) iff I |= r(π(t1 ), π(t2 ))
I, π |= ϕ1 ∧ ϕ2 iff I, π |= ϕ1 and I, π |= ϕ2
I, π |= ∃yϕ iff ∃e ∈ ∆ : π 0 = π[y/e] and I, π 0 |= ϕ
where the notation π[y/e] represents a binding π extended with π(y) = e if y is not in
the domain of π, otherwise the original value for y is replaced by e.
The validity for a TCQ Ψ and a KB K = hT , (Ai )0≤i≤n i is extended as follows:
K, i, π |= ϕ iff ∀I |=T Ai .I, π |= ϕ
K, i, π |= Ψ1 ∧ Ψ2 iff K, i, π |= Ψ1 and K, i, π |= Ψ2
K, i, π |= #Ψ iff i < n and K, i + 1, π |= Ψ
K, i, π |= #− Ψ iff i > 0 and K, i − 1, π |= Ψ
K, i, π |= Ψ1 U Ψ2 iff ∃k, i ≤ k ≤ n : K, k, π |= Ψ2
and ∀j, i ≤ j < k : K, j, π |= Ψ1
K, i, π |= Ψ1 S Ψ2 iff ∃k, 0 ≤ k ≤ i : K, k, π |= Ψ2
and ∀j, k < j ≤ i : K, j, π |= Ψ1
Answering Conjunctive Queries over a Temporally-Ordered ... 83
For a binding π, if, for every i, ∀I |=T Ai .I |= K, this implies I |= Ψ . If such
evaluation exists, we write K |= Ψ and we say π is a match for Ψ in K. For a tuple
of individuals c1 , . . . , cn mapped to a tuple of answer variables x1 , . . . , xn we say
c1 , . . . , cn is a certain answer for Ψ in K, iff K |= Ψ [x1 , . . . , xn /c1 , . . . , cn ]. We denote
a set of certain answers for Ψ as Ans(Ψ ).
Definition 5. A union of temporal conjunctive queries (UTCQ) Q(x) is a disjunction
of temporal conjunctive queries (see Definition 3):
Q(x) = {x | tconj1 (x, c) ∨ . . . ∨ tconjn (x, c)}
4 Answering a Union of Temporal CQs Over a Sequence of
ABoxes
4.1 Algorithm Answering a Union of Temporal CQs
The idea of answering a UTCQ against a set of ABoxes is to use temporal operators as
the means of detection of time points at which atemporal CQs should be evaluated. Due
to the recursive nature of such temporal operators as S , U we have to store all the
ABoxes and the values of particular CQs depending on the operator. Intuitively, given
Ψ = #φ at a time point i, φ is evaluated at the time i + 1, and so on.
To be able to combine certain answers obtained from different CQs of one TCQ,
let’s take a closer look at the nature of certain answers.
A certain answer to a CQ φ is a binding π of each xi ∈ x (distinguished variables)
to some individual name that appeared in the KB K, such that in all models of K,
K |= φ(π(x)). There could be more than one certain answer for a CQ φ, so further we
shall consider a set of certain answers for a query φ(x). A correspondent set of matches
for φ actually produces some k-ary relation, where k is the arity of a CQ φ.
A certain answer to a UCQ Φ is a combination of answers of CQs in Φ, i.e. c1 ∪. . .∪
cn where n is a number of CQs in Φ. For such a combination there are two possible
situations: (i) disjuncts φj1 , φj2 in UCQ Φ use pairwise disjoint sets of distinguished
variables (i.e. there are no common distinguished variables in two arbitrary disjuncts
of Φ); (ii) some disjuncts can share (some) distinguished variables of each other. To
deal with sets of certain answers (that are actually relations) we adopt two operators of
relational algebra, namely, × - a cross-product, and ./ - a natural join.
Cross-product operator × [5] is used for the case (i).
Definition 6. Given two bindings π1 : (x1 , . . . , xn ) → ∆, π2 : (y1 , . . . , ym ) → ∆,
their cross-product, π1 × π2 is a binding π : X → ∆ where x, y are free variables that
do not have any variables in common, and X = (x1 , . . . , xn , y1 , . . . , ym ).
Join operator ./ [5] is used for the case (ii) to join two bindings w.r.t. common
variables in both bindings are mapped to same constant.
Definition 7. Given two bindings π1 : (x1 , . . . , xn , z) → ∆, π2 : (y1 , . . . , ym , z) →
∆, their join, π1 ./ π2 is a binding π : X → ∆ where x, y, z are free variables and
X = (x1 , . . . , xn , y1 , . . . , ym , z), iff every common variable z must be mapped to same
constant c ∈ ∆.
84 N. G. Keberle
A correspondent binding for Φ will be: for (i) π = πφ1 × . . . × πφn , and for (ii)
π = πφ1 ./ . . . ./ πφn
The following theorems show applications of × and ./ for bindings.
Theorem 1. Given a formula Φ = φ1 ∧ φ2 , where φ1 , φ2 are CQ formulas, a binding
π = π1 ./ π2 is a match for Φ iff bindings π1 , π2 are matches for φ1 , φ2 .
Proof. It is true based on the definition of the join operator.
Theorem 2. Given a formula Φ = φ1 ∨ φ2 , where φ1 , φ2 are CQ formulas, a binding
π = π1 × π2 is a match for Φ iff the binding π1 is a match for φ1 or the binding π2 is a
match for φ2 .
Proof. The ⇒ direction is trivial.
For ⇐ direction, assume π1 : (x1 , . . . , xn , z) 7→ ∆, π2 : (y1 , . . . , ym , z) 7→ ∆, and
they are matches for φ1 and φ2 . From the nature of disjuntion, we know that formula Φ
is satisfiable if either φ1 or φ2 is satisfiable. That means if there is a match for either φ1
or φ2 . If z appears in both of the CQs, renaming z in one of the CQs does not change
the validity. Therefore, we have that π : (x1 , . . . , xn , z, y1 , . . . , ym , z 0 ) 7→ ∆ which is
obtained from π1 × π2 is indeed a match for Φ.
Now, consider a structure of a certain answer to a union of temporal CQs (UTCQ).
It is a combination of answers to a (set of) TCQ obtained at proper time points, referred
by temporal operators used in a UTCQ.
One more thing to be explicitly addressed is that known algorithms for conjunctive
query answering, such as [6], [7], are focused on query entailment, that is, a Boolean
conjunctuve query answering. This means that the task of answering an atemporal CQ
requires a preprocessing step, and considers a Boolean conjunctive query answering
algorithm as a black box. Namely, at the preprocessing step a candidate match (a tuple
of variables, substituted via some binding π with a tuple of individuals c) is submitted
to a Boolean conjunctive query answering engine, and that engine decides if such a
candidate match is a certain answer.
Now, present the algorithm informally.
Eliminate temporal operators in a UTCQ The important step in our algorithm is to
get a normal form where the temporal operators are used to decide at which time
point should each CQ be evaluated. This is done by iterative applecation of the
expansion rules Table 1. For every # and #− operators, we just shift one point
forward and backward. By doing these, we obtain a query that is in normal form
whose atoms are UCQs, except some recursion atom which is a TCQ.
Replace the boolean operators with relational operators Every conjunction is replaced
with join and every disjunction - with cross-product.
Retrieve an answer Use an arbitrary query answering algorithm [6–9] as a black-box
approach to compute a set of answers for a given UCQ. If the original UTCQ con-
tains U , S , 2, 2− , 3, 3− , the normal form of the transformed query might
contain a recursion. In such case, if the time point i < 0 or i > n, then return ∅,
else evaluate CQs with leading # or #− for U , S and for derived modalities (if
any).
Answering Conjunctive Queries over a Temporally-Ordered ... 85
Algorithm 1 Decide Q
Input: K = {T , (Ai )0≤i≤n } : knowledge base consists of a TBox and a sequence of ABoxes
at a time point i, 0 ≤ i ≤ n
Q : a UTCQ
Output: Ans(Q, i) - a set of certain answers to Q at time point i
Ans(Q, i) = ∅
repeat
Ans0 = Ans(Q, i)
if Q = T CQ1 ∨ T CQ2 then
Ans(Q, i) = Ans(T CQ1 , i) × Ans(T CQ2 , i)
end if
if Q = T CQ1 ∧ T CQ2 then
Ans(Q, i) = Ans(T CQ1 , i) ./ Ans(T CQ2 , i)
end if
if Q = #− T CQ then
if i=1 then
Ans(Q, i) = ∅
else
Ans(Q, i) = Ans(T CQ, i − 1)
end if
end if
if Q = #T CQ then
if i=n then
Ans(Q, i) = ∅
else
Ans(Q, i) = Ans(T CQ, i + 1)
end if
end if
if Q = T CQ1 U T CQ2 then
if i=n then
Ans(Q, i) = Ans(T CQ2 , i)
else
Ans(Q, i) = Ans(T CQ2 , i) × (Ans(T CQ1 , i) ./ Ans(Q, i + 1))
end if
end if
if Q = T CQ1 S T CQ2 then
if i=1 then
Ans(Q, i) = Ans(T CQ2 , i)
else
Ans(Q, i) = Ans(T CQ2 , i) × (Ans(T CQ1 , i) ./ Ans(Q, i − 1))
end if
end if
until Ans0 = Ans(Q, i)
return Ans(Q, i)
86 N. G. Keberle
Table 1. Equivalence rules of LTL for future operators. Taken from [4]
idempotent rule 2Ψ ≡ 22Ψ
3Ψ ≡ 33Ψ
Ψ1 U (Ψ1 U Ψ2 ) ≡ Ψ1 U Ψ2
(Ψ1 U Ψ2 ) U Ψ2 ≡ Ψ1 U Ψ2
commutativity rule 2 # Ψ ≡ #2Ψ
3 # Ψ ≡ #3Ψ
#(Ψ1 U Ψ2 ) ≡ (#Ψ1 U # Ψ2 )
distributivity rule 2(Ψ1 ∧ Ψ2 ) ≡ (2Ψ1 ∧ 2Ψ2 )
3(Ψ1 ∨ Ψ2 ) ≡ (3Ψ1 ∨ 3Ψ2 )
#(Ψ1 ∧ Ψ2 ) ≡ (#Ψ1 ∧ #Ψ2 )
#(Ψ1 ∨ Ψ2 ) ≡ (#Ψ1 ∨ #Ψ2 )
((Ψ1 ∧ Ψ2 ) U Ψ3 ) ≡ ((Ψ1 U Ψ3 ) ∧ (Ψ2 U Ψ3 ))
(Ψ1 U (Ψ2 ∨ Ψ3 )) ≡ ((Ψ1 U Ψ2 ) ∨ (Ψ1 U Ψ3 ))
temporal recursion rule 2Ψ ≡ Ψ ∧ #2Ψ
3Ψ ≡ Ψ ∨ #3Ψ
Ψ1 U Ψ2 ≡ Ψ2 ∨ (Ψ1 ∧ #(Ψ1 U Ψ2 ))
absorption rule 323Ψ ≡ 23Ψ
232Ψ ≡ 32Ψ
In Table 1, presented are some equivalence rules in LTL, used in Algorithm 1.
For the illustration of Algorithm 1 consider some examples, assuming that Algo-
rithm 1 returns a set Ans of answers to Ψ at the time point i.
Example 3. Given a TCQ query Ψ = #− (Φ1 U Φ2 ) at a point i.
Ans(Ψ, i) = Ans #− (Φ1 U Φ2 ), i
∗\move back one pointby #−
= Ans Φ1 U Φ2 , i − 1
∗\expansion rule for U
= Ans Φ2 ∨ (Φ1 ∧ #Ψ ), i − 1
∗\transforming ∨
= Ans Φ2 , i − 1 × Ans Φ1 ∧ #Ψ, i − 1
∗\transforming ∧
= Ans Φ2 , i − 1 × Ans Φ1 , i − 1 ./ Ans # Ψ, i − 1
∗\move forward one point by #
= Ans Φ2 , i − 1 × Ans Φ1 , i − 1 ./ Ans Ψ, i
If i = 0 in Ans Φ2 , i − 1 and Ans Φ1 , i − 1 , then the evaluation of Ψ is the empty
set.
A more complex example is given below.
Answering Conjunctive Queries over a Temporally-Ordered ... 87
Example 4. Given a TCQ query Ψ = 3− (Φ1 U Φ2 ) at a point i.
Ans(Ψ, i) = Ans 3− (Φ1 U Φ2 ), i
∗\expansion rule for 3−
= Ans (Φ1 U Φ2 ) ∨ #− Ψ, i
∗\transforming ∨
= Ans Φ1 U Φ2 , i × Ans #− Ψ, i
∗\move back one point
by #−
= Ans Φ1 U Φ2 , i × Ans Ψ, i − 1
∗\We substitute Φ1 U Φ2 with Ψ 0
Expansion rule for U
= Ans Φ2 ∨ (Φ1 ∧ #Ψ 0 , i × Ans Ψ, i − 1
∗\transforming ∨
= Ans Φ2 , i × Ans Φ1 ∧ #Ψ 0 , i × Ans Ψ, i − 1
∗\transforming ∧
=
Ans Φ 2 , i ×
Ans Φ1 , i ./ Ans # Ψ 0 , i × Ans Ψ, i − 1
∗\move forward one point by #
= Ans Φ2 , i × Ans Φ1 , i ./ Ans Ψ 0 , i + 1 × Ans Ψ, i − 1
If i = n in Ans Ψ 0 , i + 1 , then Ans Ψ 0 , i + 1 is evaluated to the empty set.
There is one thing we have to ensure that in the intersection of two sets of answers
for conjunction of CQs a certain answer is obtained, i.e. there is a common answer for
both CQs, otherwise an empty set. One way to do this is to retrieve all answers for each
CQ and then to intersect them to get some common answers. Another way is first to
retrieve an answer of a UCQ and then to decide if this answer is also the answer for
the other CQs in the conjunction, otherwise keep retrieving and deciding until there is
no more answer obtained. The former way is preferred since it offers more practical
solution. It means that we can deal with it using relational algebra operators or database
language operators.
4.2 Termination, Soundness, Completeness of the Algorithm
Definition 8. (UTCQ closure). Given a temporal union of conjunctive queries Q, its
closure set, Cl(Q) is a set of query atoms closed under the following rules
if q ∈ Q then q ∈ Cl(Q)
if #− q ∈ Q then q ∈ Cl(Q)
if #q ∈ Q then q ∈ Cl(Q)
if q1 ∧ q2 then q1 , q2 ∈ Cl(Q)
if q1 ∨ q2 then q1 , q2 ∈ Cl(Q)
if q1 U q2 then q1 , q2 , #(q1 U q2 ) ∈ Cl(Q)
if q1 S q2 then q1 , q2 , #− (q1 S q2 ) ∈ Cl(Q)
Since a closure set for a UTCQ is finite, Algorithm 1 terminates after a finite number
of steps.
88 N. G. Keberle
Theorem 3. (Local) termination. Given a UTCQ Q and a knowledge base
K = {T , (Ai )0≤i≤n }. Algorithm 1 always terminates.
Proof. We can show the local termination inductively.
Base case. Any query is also cointained in the closure set of itself.
Inductive case.
(C(a), r(a1 , a2 )) If we have a query Q which is atomic, then the closure set contains
C(a) or r(a1 , a2 ).
(#− T CQ) For such query Cl(Q) = {T CQ, #− T CQ}, i.e. evaluated are two el-
ements, and in case of i = 0 the value of #− T CQ is known to be ∅, so Algorithm 1
stops after two evaluations.
(T CQ1 U T CQ2 ) For such query Cl(Q) = {T CQ2 , T CQ1 , T CQ1 U T CQ2 ,
# (T CQ1 U T CQ2 )}
(T CQ1 S T CQ2 ) For such query Cl(Q) = {T CQ2 , T CQ1 , T CQ1 S T CQ2 ,
# (T CQ1 S T CQ2 )}
Theorem 4. Soundness. If for UTCQ Q its answer set Ans(Q(x), i), obtained with Al-
gorithm 1, is not empty, then Q has at least those certain answers that are in Ans(Q, i).
Proof. We prove by induction. We start with evaluating non-temporal query, i.e. a query
containing no temporal operator.
Base case If we have an atomic query in the form of C(a), then using any ap-
proach of CQ answering we obtain all the answers for the query Q entailment over
K = {T , (Ai )0≤i≤n }. If K |= C(a) and a ∈ Ans(Q(x), i), the function returns a
and this value is stored in Ans(Q(x), i). By Definition 4, this result tells us that the
individual a is a certain answer to the query C(x) w.r.t. the match π(x) = a. The same
result is obtained if we have atomic query in the form of r(a, b).
Inductive case can be obtained by Definition 4.
Theorem 5. Completeness. If a UTCQ Q has a certain answer ans, then Algorithm 1
shows that this answer is in Ans(Q, i).
Proof. By contradiction. Assume that (i) Q(x) has a certain answer ans w.r.t π, (ii)
Ans(Q, i) - is a set of certain answers obtained by Algorithm 1, and (iii) ans 6∈
Ans(Q, i). By (i), we know that K |= Q(ans) and that for all time points 0 ≤ i ≤ n
in all models I, such that I |= K, I |= Q(ans). By (ii), for Algorithm 1 to return
Ans(Q, i) such that ans 6∈ Ans(Q, i) there are several reasons for it.
Q is atomic. If Q is atomic, i.e. in the form C(x) or r(x, y), then we know that
Ans(Q, i) does not contain ans. This means that there is a model I of a knowledge
base K which does not entail Q(ans). But this is a contradiction to our assumption (i).
(T CQ1 ∧ T CQ2 ). If Ans(Q, i) does not contain ans, according to Algorithm 1
it means that ans 6∈ Ans(T CQ1 , i) ./ Ans(T CQ2 , i). This, in turn, leads to the
existence of a model I of a knowledge base K such that I |= T CQ1 (ans) and I 2
T CQ2 (ans) or vice versa, that contradicts to (i).
(T CQ1 ∨ T CQ2 ). If Ans(Q, i) does not contain ans, according to Algorithm 1
it means that ans 6∈ Ans(T CQ1 , i) × Ans(T CQ2 , i). This, in turn, leads to the ex-
istence of a model I of a knowledge base K such that either I 2 T CQ1 (ans) or
I 2 T CQ2 (ans), that contradicts to (i).
Answering Conjunctive Queries over a Temporally-Ordered ... 89
(#− T CQ). If Ans(Q, i) does not contain ans, according to Algorithm 1 it means
that ans 6∈ Ans(Q, i − 1). This, in turn, leads to the existence of a model I of a
knowledge base K such that I, i − 1 2 Q(ans), that contradicts to (i).
(T CQ1 U T CQ2 ). If Ans(Q, i) does not contain ans, according to Algorithm 1
it means that ans 6∈ Ans(T CQ2 , i) × (Ans(T CQ1 , i) ./ Ans(Q, i + 1). This, in
turn, leads to the existence of a model I of a knowledge base K such that either I, i 2
T CQ2 (ans) or I, i 2 T CQ1 (ans) and I, i + 1 2 Q, that contradicts to (i).
(T CQ1 S T CQ2 ). If Ans(Q, i) does not contain ans, according to Algorithm 1
it means that ans 6∈ Ans(T CQ2 , i) × (Ans(T CQ1 , i) ./ Ans(Q, i − 1). This, in
turn, leads to the existence of a model I of a knowledge base K such that either I, i 2
T CQ2 (ans) or I, i 2 T CQ1 (ans) and I, i − 1 2 Q, that contradicts to (i).
The proof for the temporal operator # acting in the direction of future can be com-
pleted in the same manner.
5 Related Work and Conclusions
Transition graphs for a temporal query language answering over a finite set of versions
of a database were investigated in [10]. The expressivity of a temporal query language
presented is however restricted either to past [11], or to future [10], [12] direction
of time. Known are several algorithms for answering unions of conjunctive queries
over knowledge bases with static TBox and ABox, for example works of Ortiz [6],
Glimm [7], Tessaris [9], Motik [8] should be mentioned. Any of those algorithms could
serve as a basis for finding answers to atemporal CQs at particular time points, whereas
possible extensions of those algorithms for the application to a sequence of ABoxes is
an open question. A language of temporal conjunctive queries with negation, together
with the computational and combined computational complexity is intriduced in [13].
Summing up, obtaining benefits from keeping a large evolving ABox of a knowledge
base in a database and applying TBox of that knowledge base to obtain missing as-
sertional axioms is one of the ways of dealing with complex evolving domains. It is
interesting, due to high computational complexity of temporal conjunctive query an-
swering in general, to find a balance between the expressivity of a query language and
its practical applicability.
Acknowledgements The presented results were obtained during the research visit of
the author to the Chair of Automata Theory at Dresden University of Technology. The
author is grateful to the group of Prof. Franz Baader, and in particular, Eldora, Marcel
Lippmann and Anni-Yasmin Turhan for the fruitful discussions and ideas at the stage
of early drafts of the paper.
References
1. Poggi, A., Lembo, D., Calvanese, D., De Giacomo, G., Lenzerini, M., Rosati R. Linking Data
to Ontologies. J. on Data Semantics, X, 133–173 (2008)
90 N. G. Keberle
2. Baader, F., Bauer, A., Baumgartner, P., Cregan,A., Gabaldon,A., Ji, K., Lee,K., Rajarat-
nam,D., Schwitter, R. A novel architecture for situation awareness systems. In: Giese, M.
and Waaler, A. (eds.) Proc. 18th International Conference on Automated Reasoning with An-
alytic Tableaux and Related Methods (Tableaux 2009). LNCS, vol. 5607, pp. 77–92. Springer,
Berlin/Heidelberg (2009)
3. Baader, F., Calvanese, D., McGuinness, D.L., Nardi, D., Patel-Schneider, P.F. (eds.). The De-
scription Logic Handbook: Theory, Implementation, and Applications. Cambridge University
Press (2003)
4. Baier, C., Katoen, J.-P. Principles of Model Checking. The MIT Press, Cambridge, Mas-
sachusetts, USA (2008)
5. Abiteboul, S., Hull, R., Vianu, V. Foundations of Databases. Addison-Wesley (1995)
6. Ortiz de la Fuente, M.M. Query Answering in Expressive Description Logics Techniques and
Complexity Results. PhD Thesis. Technischen Universitt Wien, Fakultt fr Informatik (2010)
7. Glimm, B. Querying Description Logic Knowledge Bases. PhD Thesis. The University of
Manchester (2007)
8. Motik, B. Reasoning in Description Logics using Resolution and Deductive Databases. Uni-
vesitt Karlsruhe (2006)
9. Tessaris, S. Questions and answers: reasoning and querying in Description Logic. The Uni-
versity of Manchester (2001)
10. Lipeck, U.W. Transformation of Dynamic Integrity Constraints into Transaction Specifica-
tions. Theor. Comput. Sci., 76(1), pp. 115–143 (1990)
11. Schwiderski, S., Hartmann, T., Saake, G. Monitoring Temporal Preconditions in a Behaviour
Oriented Object Model. Data Knowl. Eng., 14(2), pp. 143–186 (1994)
12. Lipeck, U.W., Feng, D. Construction of Deterministic Transition Graphs from Dynamic In-
tegrity Constraints. LNCS, vol. 344, pp.166–179. Springer Verlag (1989)
13. Baader F., Borgwardt, S., Lippmann, M. On the Complexity of Temporal Query
Answering. Technical report LTCS-Report 13-01. Available at http://lat.inf.tu-
dresden.de/research/reports/2013/BaBoLi-LTCS-13-01.pdf (2013)
An Adaptive Forecasting of Nonlinear Nonstationary
Time Series under Short Learning Samples
Elena Mantula1 and Vladimir Mashtalir1
1
Kharkiv National University of Radio Electronics, informatics department
Lenin ave., 14, 61166, Kharkiv, Ukraine
Mashtalir@kture.kharkov.ua, ElenaMantula@gmail.com
Abstract. Methods of nonstationary nonlinear time series forecasting under
bounded a priori information provide an interdisciplinary applications area that
is concerned with learning and adaptation of solutions from a traditional artifi-
cial intelligence point of view. It is extremely difficult to solve this type of
problems in its general form, therefore, an approach based on the additive
nonlinear auto regressive model with exogenous inputs and implemented on the
base of parallel adalines set has been proposed. To find optimal combination of
forecasts, an improvement of global random search has been suggested.
Keywords. Neural networks, forecasting model, combination of forecasts
Key terms. Environment, MathematicalModel
1 Introduction
‘Conscious’ decision making, in all possible varieties, is perhaps the most principal
goal of artificial intelligence systems. Necessary ‘creativity’ implies the ability to
produce novel solutions which are better than previous ones. The computational tools
that assist in decision making should be such that they should take into all aspects of
dissimilarity between a priori and a posteriori uncertainty. Uncertainty account is, per
se, a manifestation of information deficiency, and relevant information is, on the con-
trary, a capacity to reduce uncertainty. An elimination of such rich in content gaps
provides groundwork of knowledge engineering and management. In machine intelli-
gence, manifold forecasts can be used for knowledge producing. The goal of the paper
consists in reasonable (perfectly optimal) combination of forecasts to provide reliable
semantic interpretation of achieved results with purpose knowledge generation.
Nowadays mathematical forecasting models of the behavior of objects, systems
and phenomena in a wide variety of applications are well understood. There is a
wealth of publications on this subject. It should be noted that the behavior of the ob-
jects is often given in the form of time series. Thus to forecast its behavior a variety of
approaches to the analysis of time series can be used. Such approaches can be either
traditional statistical methods (regression, correlation, spectral, Box-Jenkins) or adap-
tive, based on an exponential smoothing, tuning or learning forecasting models, or
92 E. Mantula and V. Mashtalir
intellectual, using vrious neural networks.
At present there are many objects (financial, economical, biomedical, etc.), de-
scribed by time series containing unknown behavior trends, seasonal components,
stochastic and random components, which significantly complicate synthesis of an
effective predictive model. This complexity is especially pronounced in the environ-
mental monitoring problems [1], where the analyzing time series have in equal meas-
ure stochastic and chaotic type of changes, have apparent nonstationarity and are sub-
jected to striking changes.
In these conditions artificial ccc have proved to be useful tools in the best way [2-
13]. As a rule, they realize so-called NARX-model [14], which has the form
ŷ(k ) f (y(k 1),...,y(k nA ), x(k 1),..., x(k nB ) (1)
where ŷ(k ) is an estimation of forecasted variable y(k ) at discrete time k 1,2,...;
f () denotes certain nonlinear transform which is realized by a neural network; x(k )
is the observed exogenous factor that influences the behavior of y(k ) ; nA , nB are
observations memory parameters.
Moreover, it is not a matter of available observations insufficiency, since proper-
ties of time series (e.g. such indicator as air pollution in ecological forecasting) are
changed so often that a neural network does not have time to detect separate station-
ary parts. In this connection there is a need to construct based on the neural network
approach simplified predictive models for training which require the small enough
volume data set.
2 Synthesis of a forecasting model
In conditions of input data lack instead of NARX-model (1) it is appropriate to use the
so-called ANARX-model introduced in [15, 16] and fully investigated in [17, 18]. In
general ANARX-model can be written as
ŷ(k ) f1(y(k 1), x(k 1)) f 2 (y(k 2), x(k 2)) ...
f max {nA , nB}(y(k nA ), x(k nB )) (2)
max {n A , nB }
f l (y(k l ), x(k l ))
l 1
where original task is decomposed into many local ones with two input variables
y(k l ), x(k l ) , l 1,2,..., max {nA ,nB} .
For such nonlinear transforms it is quite convenient to use so-called N-adaline
(abbr.: adaptive linear element) [19-21] that provide quadratic approximation of the
data sequence. Fig. 1(a) demonstrates the architecture of N-adaline and (b) illustrates
the architecture of ANARX-model constructed using N-adaline.
As we can see, N-adaline represents a generally accepted two-input adaline with a
nonlinear preprocessor formed by three blocks of the product ( ) and the evaluator
of the quadratic combination in the form
f l (y(k l ), x(k l )) wl 0 wl1 y(k l ) wl 2 y 2 (k l ) wl 3 y(k l )x(k l ) wl 4 x 2 (k l ) wl 5 x(k l )
An Adaptive Forecasting of Nonlinear Nonstationary Time Series … 93
where each N-adaline contains 6 synaptic weights wlp , l 1,2,..., max {n A , nB } ,
p 0,1,...,5 . As a matter of fact, ANARX-model is formed by two lines of delay ele-
ments z 1 and max {n A ,nB } parallel learned N-adaline.
(a) N-adaline (b) ANARX models
Fig. 1. N-adaline and ANARX models based on N-adalines
Each from N-adalines is configured with any of the linear learning algorithms [22],
however, it is clear that a limited amount of a priori information requires the use of
time-optimal procedures. As such can be, for example, adaptive-multiplicative modi-
fication of Kachmarz adaptive algorithm [23], which assumes in this case the form
y(k ) wlT (k 1)l (k )
wl (k ) wl (k 1) 2
l (k ) (3)
l (k )
where wl (wl 0 ,wl1,wl 2 ,wl 3 ,wl 4 ,wl 5 )T ; l (k ) (1, y(k l ), y 2 (k l ), y(k l )x(k l ),
x 2 (k l ), x(k l ))T ; 0 2 , 0 are some algorithm parameters selected on the
base of empirical reasons.
If the data sequences are ‘contaminated’ by perturbations, instead of the one-step
algorithm (3) it is profitably to apply procedures that provide filtering of perturbation
and at the same time they have to be suitable for using in non-stationary conditions. It
should be noted that modification of the recursive least squares method on a sliding
window can be used [24]. The traditional estimation method of least squares on the
window with s observations has the form
k 1 k
wl (k ) ( k s 1 l ()l ()) k s 1 l ()y()
T
and recurrent one can be presented as
94 E. Mantula and V. Mashtalir
Ps (k 1)l (k )Tl (k )Ps (k 1)
P(k ) Ps (k 1) ,
1 Tl ( k )Ps ( k 1 )l ( k )
P(k )l (k s)lT (k s)P(k )
Ps (k ) P(k ) , (4)
1 l (k s)P(k )l (k s )
ps (k ) ps (k 1) l (k )y(k ) l (k s)y(k s),
wl (k ) Ps (k )ps (k ).
We also note that if the algorithm (3) is in fact time-optimal gradient procedure,
then the algorithm (4) is produced by Gaussian-Newton optimization procedure.
3 Optimal combination of forecasts
In real conditions the choice of the forecasting model structure is not a trivial task,
especially that the same time series can be effectively described by a variety of differ-
ent models. Also, the value of the lag orders nA , nB remains unknown what makes it
necessary to consider a set of competing models, and nonstationarity of analyzed
series necessitates the use of various learning algorithms (in this case, (3), (4)) with
different values , ,s . Thus, there arises a set of forecasts of the same process, from
which we have to select the best.
To find the best forecast it is possible to use sufficiently effective approach, based
on the optimal combination of forecasts [25], under which optimal in the sense of
given criterion J c linear combination is searching for a set of existing forecasts of
the same series ŷ j (k ), j 1,2,..., m
m
ˆy(k ) c j ˆy j (k ) (5)
j 1
where the parameters of the combination satisfy the condition of unbiasedness
m
c j 1. (6)
j 1
In [25], an analytical approach to the weights c j finding in (5) by optimizing the
sum of squared errors criterion for forecasting with the constraints (6) is proposed.
The use of one-step squared forecast errors criterion leads to the estimation
ŷ j (k )
c j (k ) m .
j ŷ (k )
j 1
However, combining of the analytical parameter estimates can be obtained under
application of standard quadratic criterion J c solely that specified by linearity of it
derivatives so the solution of the problem reduces to solving a system of linear equa-
tions. At the same time for practitioners as a rule assess of the quality of forecasting
An Adaptive Forecasting of Nonlinear Nonstationary Time Series … 95
using the residual variance is unconvincing, and therefore characteristics allowing to
estimate the accuracy in percentage are generally used, such as the criterion of a
minimum of absolute percentage error
N y(k ) ˆy(k )
MAPE 100% (7)
k 1 y(k )
or maximum of the determination coefficient
N
(y( k ) ˆy( k ))2
R 2 (1 N k 1 N
)100% . (8)
1 2
(y( k ) y(k ))
k 1 N k 1
It is obvious that in this case analytical estimations can not be obtained, and the use
of gradient optimization procedures becomes more complicated due to sufficiently
complex properties of functions (7), (8). In this connection the use of genetic algo-
rithms is proposed in [26, 27]. Though such algorithms can find the global extremum,
their own distinctive features are numerical awkwardness, they have a set of free pa-
rameters necessary defined by the user and at last it should be mentioned a low rate of
convergence. Therefore, notice should be taken to more an efficient approach based
on the random search [28] and its adaptive modifications. The most simple procedure,
which allows to search for a global extremum, is walking random global search [28].
In general, this procedure is a statistic extension of the regular gradient search, and to
provide the global search, random disturbance (k ) superimposes on character on a
gradient movement what creates stochastic walking mode.
In the continuous case, the gradient method of minimization (maximization) of the
goal function J c (t ) is reduced to the motion of a point c(t ) c1(t ),..., c j (t ),..., cm (t )
in m -dimensional space of adjustable parameters by a force directed toward the anti-
gradient.
The trajectory of movement by antigradient c(t ) leads tuning process to a singular
point. If starting point c(0) belongs to an attraction region of global extremum then
the corresponding trajectory will lead to a global minimum of the function J c (t ) . But
if the point c(0) does not belong to this region, the movement in the direction of anti-
gradient will result in a local minimum, from which it is impossible to get out under
the influence of forces directed by antigradient. Exactly because, it is helpful to use a
random mechanism. Random shocks may help point c(t ) to overcome the barrier that
separates the local minimum in which the learning process hit from the area in which
the objective function J c(t ) could further decrease. Under the influence of ‘skew’
toward anti-gradient and random shocks such movement is determined by the differ-
ential equation
dc(t )
c J c (t ) (t )
dt
where (t ) is m -dimensional normal random process with zero mathematical expec-
96 E. Mantula and V. Mashtalir
tation, delta-figurative autocorrelation function and components variance 2 ; is
parameter of step, c denotes gradient vector. It should be emphasized that for func-
tion (7) the components of the gradient can acquire the value 1 or 1 . Generally,
this algorithm provides searching for a global extremum [29].
Searching for global extremum can be speed up by reasonable selection of 2 and
an adaptation during this process can be introduced in two ways. First, under intro-
ducing inertia in the learning process, it is possible to get a search similar to the
movement by the method of ‘heavy ball’ [30]. Such movement is described by the
differential equation
d 2c(t ) dc(t )
2
b J c (t ) (t ) (9)
dt dt
where b is shockproofing coefficient (the more b , the less manifest of inserted iner-
tia).
On time series processing, i.e. in discrete time, procedure (9) corresponds to the
learning algorithm, described by the second order difference equation [31]
c(k ) c(k 1) bc(k 2) (k )c J c (k ) (t ) (10)
coinciding under b 0 with walking random search. It is interesting to note that (10)
is none other than the ARX- model of the second-order.
Second, the adaptation in the process of global search can be introduced by random
process (t ) control, for example,
d (t ) dJ c (t )
(t ) 2 H(t ) (11)
dt dt
where 0 is a autocorrelation parameter of random process (t ) ; H(t ) is a vector
of flat random noise. Introduce a modification of (11) in the discrete form
(k ) (1 )(k 1) (k ) J c (k ) 2 H(k ) (12)
where is the symbol of the first difference (discrete analogue of the derivative).
As it is easily seen from (11), (12), the optimization of the search process can be
performed by appropriate selection of parameters , and 2 , since each of them
acts on the certain properties of the search. Indeed, variation of the autocorrelation
parameter determines the rate of the process (k ) decay that regulates its relations
with the past. Thus, one can have an influence upon a search making it more or less
dependent on the previous history if it is necessary.
Some few words of comment are desirable for parameters and interaction
explanation. If the search step determines the intensity of accumulation of learn-
ing experience, then characterizes the level of this experience forgetting during the
search. In this sense, these parameters are antagonistic. If in general 0 and there
is no forgetting the vector (k ) increases in the direction of anti-gradient. Variance
An Adaptive Forecasting of Nonlinear Nonstationary Time Series … 97
of the process (k ) is determined by the value 2 and intensity of the flat random
noise disturbance H(k ) . If 2 is sufficiently large then search may become unstable
and, at low value, global properties are worsening. Thus, the use of a modified global
random search allows simplify significantly the process of linear combination
cj (k ), j 1,2,..., m tuning.
4 Conclusion
The problem of nonstationary nonlinear time series forecasting under bounded a priori
information has been considered. An approach based on the additive nonlinear auto
regressive model with exogenous inputs and implemented on the base of parallel
adalines set has been proposed. To find optimal combination of forecasts, an im-
provement of global random search has been suggested. Distinctive feature of the
approach is the computational simplicity and high performance attained by significant
reducing the number of adjustable parameters.
References
1. Zanetti, P.: Air Pollution Modelling. Van Nostrand Reinhold, New York (1990)
2. Reich, S.L., Gomez, D.R., Dawidowski, L.E.: Artificial Neural Network for the Identifica-
tion of Unknown Air Pollution Sources. Atmosphere Environment, Vol. 33, pp. 3045-3052
(1999)
3. Perez, P., Trier, A., Reyes, J.: Prediction of PM2.5 Concentration Several Hours in Ad-
vance Using Neural Networks in Santiago, Chile. Atmospheric Environmental, Vol. 34,
pp. 1189–1196 (2000)
4. Niska, N., Hiltunen, T., Karppinen, A., Ruuskanen, J., Kolehmanen, M.: Evolving the
Neural Network Model for Forecasting Air Pollution Time Series. Engineering Applica-
tion of Artificial Intelligence, Vol. 17, 159–167 (2004)
5. Corani G.: Air Quality Prediction in Milan: Feed-Forward Neural Networks, Pruned Neu-
ral Networks and Lazy Learning. Ecological Modeling, Vol. 185, pp. 513–529 (2005)
6. Athanasiadis, I.N., Karatzas, K.D., Mitkas, P.A.: Classification Techniques for Air Quality
Forecasting. In: Brewka G., Coradeschi S., Perini A. and Traverso P. (eds.): Proc. 17th
European Conference on Artificial Intelligence, IOS Press, Amsterdam, 4.1–4.7 (2006)
7. Perez, P., Reyes, J.: An Integrated Neural Network Model for PM10 Forecasting. Atmos-
pheric Environment. Vol. 40, pp. 2845–2857 (2006)
8. Lira, T.S., Barrozo, M.A.S., Assis. A.J.: Air Quality Prediction in Uberlandia, Brasil, Us-
ing Linear Models and Neural Networks. In: Plesu V., Agachi P. (eds.): Proc. 17th Euro-
pean Symp. on Computer Aided Process Engineering, Elsevier, Amsterdam, pp. 1–6
(2007)
9. Kurt, A., Gulbagci, B., Karaca, F., Alagha, O.: An Online Air Pollution Forecasting Sys-
tem Using Neural Networks. Environmental International, Vol. 34 (2008) 592–598
10. Carnevale, C., Finzi, G., Pisoni, E., Volta, M.: Neuro-Fuzzy and Neural Network Systems
for Air Quality Control. Atmospheric Environmental, Vol. 43, pp. 4811–4821 (2009)
98 E. Mantula and V. Mashtalir
11. Nagendra, S.M., Shiva, Khare M.: Modelling Urban Air Quality Using Artificial Neural
Network. Clean Technical Environmental Policy, Vol. 7, pp. 116–126 (2005)
12. Aktan, M, Bayraktar, H.: The Neural Network Modeling of Suspended Particulate Matter
with Autoregressive Structure. Ekoloji, Vol. 19, No. 74, pp. 32–37 (2010)
13. Esau, I.: On Application of Artificial Neural Network Methods in Large-Eddy Simulations
with Unresolved Urban Surfaces. Modern Applied Science, Vol. 4, No. 8, 3–11 (2010)
14. Nelles, O.: Nonlinear System Identification: From Classical Approaches to Neural Net-
works and Fuzzy Models. Springer, Berlin (2001)
15. Chowdhury, F.N., Input-Output Modeling of Nonlinear Systems with Time-Varying Lin-
ear Models. IEEE Trans. on Automatic Control, Vol. 45, No. 7, pp. 1355–1358 (2000)
16. Kotta, Ü., Sadegh, N.: Two Approaches for State Space Realization of NARMA Models:
Bridging the Gap. Mathematical and Computer Modeling of Dynamical Systems, Vol. 8,
No. 1, pp. 21–32 (2002)
17. Belikov, J., Vassiljeva, K., Petlenkov, E., Nomm S.: A Novel Taylor Series Based Ap-
proach for Control Computation in NN–ANARX Structure Based Control of Nonlinear
Systems. In: Proc. 27th Chinese Control Conference, Beihang University Press, Kunming,
pp. 474–478 (2008)
18. Vassiljeva, K., Petlenkov, E., Belikov, J.: State-Space Control of Nonlinear Systems Iden-
tified by ANARX and Neural Network Based SANARX Models. In: Proc. WCCI 2010
IEEE World Congress on Computational Intelligence, IEEE CSS, Piscataway, pp. 3816–
3823 (2010)
19. Pham, D.T. Liu, X.: Modeling and Prediction Using GMDH Networks of Adalines with
Nonlinear Preprocessors. Int. J. System Science Vol. 25, No. 11 (1994) 1743–1759
20. Pham, D.T. Liu, X.: Neural Networks for Identification, Prediction and Control. Springer,
London (1995)
21. Rudenko, O.G., Bodyanskiy, Ie.V.: Artificial Neural Networks. SMIT, Kharkov (2005) (in
Russian)
22. Haykyn, S.: Neural Networks: A Comprehensive Foundation. Prentice Hall. Inc., New
York (1999)
23. Raybman, N.S., Chadeev, V.M.: Creating of Manufacture Process Models. Energiya, Mos-
cow (1975) (in Russian)
24. Perelman, I.I.: Operative Identification of Control Objects. Energoatomizdat, Moscow
(1982) (in Russian)
25. Sharkeya, A.J.C.: On Combining Artificial Neural Nets. Connection Science, Vol. 8, No.
3, pp. 299–314 (1996)
26. Zagoryjko, N.G.: Empirical Prediction. Nauka, Novosibirsk (1979) (in Russian)
27. Zagoryjko, N.G.: Applied Approach of Data Analysis, р. 264 (1999) (in Russian)
28. Rastrigin, L.A.: Statistical Search Technology. Nauka, Moscow (1968) (in Russian)
29. Rastrigin, L.A.: Systems of Extremal Control. Nauka, Moscow (1974) (in Russian)
30. Polyak B.T.: Introduction into Optimization. Nauka, Moscow (1983) (in Russian)
31. Bodyanskiy, Ie. V., Rudenko, O.G.: Artificial Neural Networks: Arhitectures, Learning,
Applications. TELETEX, Kharkov (2004) (in Russian)
Application of an Instance Migration Solution
to Industrial Ontologies
Maxim Davidovsky1, Vadim Ermolayev1 and Vyacheslav Tolok 2
1
Department of IT, Zaporozhye National University,
66 Zhukovskogo st., 69600 Zaporozhye, Ukraine
m.davidovsky@gmail.com, vadim@ermolayev.com
2
Department of Mathematical Modeling, Zaporozhye National University,
66 Zhukovskogo st., 69600 Zaporozhye, Ukraine
vyacheslav-tolok@yandex.ru
Abstract. The paper presents the results of evaluating the software solution for
ontology instance migration problem in the use case involving the ontologies
used in construction industry – freeClass and eClassOWL with the Bau-
DataWeb dataset representing the individuals. Ontology instance migration
problem is understood as a sub-problem of ontology alignment. Our methodol-
ogy assumes (semi-) automated iterative process possibly involving a human
for validating the results. The process consists of the two steps: (1) schema-
based mappings discovery done by the agent-based matcher software; and (2)
ontology instance transformation and migration according to the discovered
mappings done by the ontology instance migration engine software. The
evaluation experiment has been conducted in two phases and yielded results of
acceptable quality in terms of precision, recall, and f-measure.
Keywords. Ontology alignment, industrial application, ontology instance mi-
gration, evaluation experiment.
Key terms. Industry, Integration, Interoperability, KnowledgeManagement-
Process, AgentBasedSystem,
1 Introduction
Ontologies are being widely adopted today in the academic world and increasingly
attract the attention of researchers and practitioners in information technology and
knowledge-based system development and applications. Many authors, e.g. [1], argue
that ontologies constitute the substance of the advanced technologies for solving the
problems of interoperability, communication, and cooperation between different ap-
plications within the same environment. Indeed, ontologies conceptualize semantics
of the domains within a discourse that are common for interoperating systems. Thus,
100 M. Davidovsky, V. Ermolayev and V. Tolok
ontologies serve as a bridge for “understanding” between the systems or their parts.
Despite that, application of ontologies in industry still faces several problems.
The first group of problems concerns the inertia that is typical for the process of
application of advanced technologies in industry, e. g. [2]. This paper reflects the
views of the practitioners who have witnessed incomprehension and opposition in
trying to solve customer problems using ontologies. These problems are attempted to
be resolved through establishing a closer contact with domain knowledge stake-
holders and their more active involvement in the development of ontologies – e.g. [3].
Another complementary and important activity is lowering the effort for developing
ontologies which could be done via providing the tool support for domain experts
taking part in ontology development.
The other important stratum of problems in the application of ontologies in indus-
try is related to the re-use of existing large industrial knowledge bases, collections, or
ontologies and the exploitation of those knowledge assets within large enterprise in-
formation systems (IS). Obviously it is obligatory to provide stable interoperation of
ISs in industrial settings to prevent substantial errors in maintenance, production, and
sales. However the use of ontologies per se doesn’t completely solve interoperability
issues as it essentially raises heterogeneity problems to a higher level [4]. So, the
methods for aligning ontologies need to be provided to understand and explicitly
specify semantic mappings between these different conceptualisations. Industrial
ontologies as a rule contain large quantity of individuals (or instances). Hence, an
important and typical sub-problem of ontology alignment in industrial settings is on-
tology instance migration that is the process of transferring instances between aligned
ontologies. The numbers of the individuals in industrial knowledge bases is very often
high, so their manual alignment is not feasible. Therefore it is important to provide
the tools that at least partially automate the process of alignment and do that with the
quality acceptable for industries. Another important aspect of the use of ontologies in
industrial settings is that industrial ISs are often distributed and belong to autonomous
business entities. In such settings using intelligent software agents for ontology
alignment and ontology instance migration in particular becomes an attractive imple-
mentation pathway.
The remainder of the paper is organized as follows. In section 2 we provide a clas-
sification of industrial applications of ontology alignment types of problems and de-
scribe some typical use cases. Based on this classification, we analyze industrial re-
quirements to ontology alignment solutions. Section 3 outlines our software solution
for ontology instance migration problem. Section 4 reports about the setup and results
of our evaluation experiments. Finally the conclusion is given and the plans for the
future work are outlined.
2 Related Work, Applications, and Use Cases
Surveys of ontology alignment for a wide range of applications can be found in [5],
[6], [7]. Applications of agent-based ontology alignment and respective requirements
Application of an Instance Migration Solution … 101
are analyzed in [8]. This paper focuses on industrial applications of ontology
alignment in broad and ontology instance migration as its sub-problem [8].
The following industrial application categories may be outlined that require
ontology alignment and instance migration solutions.
1. Industrial knowledge-driven simulation models. Simulation models are
widely used in industry ([9], [10], [11]). The complexity level of modern simulation
systems requires the use of knowledge-based models. This knowledge may be related
to various branches of science, engineering disciplines, can contain different models
satisfying different demands. This requires the use of ontologies and related activities
such as ontology merging and alignment.
2. Industrial information systems in the context of Semantic Web and
eCommerce. eCommerce is a type of industry where buying and selling of product or
service is conducted over electronic systems such as the Internet and other computer
networks. In order to perform such an exchange of business information, this informa-
tion must contain product (or service) descriptions. As a rule such information is pre-
sented in the form of product or service ontology [12]. Good examples of such on-
tologies are [13], [14], [15]. When a business process involves more than one party or
in a case of using more than one source respective ontologies obviously have to be
aligned. This situation is also typical for The Semantic Web where ontologies along
with intelligent software agents are the main pillars [16].
3. Integration and interoperability of heterogeneous enterprise ISs. Today
information ecosystem of a modern enterprise as a rule contains numbers of applica-
tions from different vendors and used for different purposes. In order to effectively
use these heterogeneous applications together with distributed data and knowledge
repositories they must be integrated into a single system. Likewise implementation
and deployment of new software solutions must be reconciled and integrated with
legacy software systems. Here ontologies may be used not only as domain knowledge
representation models, but also as mediators for integration of heterogeneous applica-
tions. Enterprise integration attracts substantial interest of research community and a
number of solutions are proposed (e.g., [17], [18]).
4. Knowledge sharing and migration between enterprise ISs. Interaction and
cooperation of modern enterprises often implies knowledge sharing and migration. In
such a way enterprise may enrich and harmonize their knowledge assets. In this case
knowledge models obviously must be reconciled and aligned. This issue is not widely
addressed in literature (but some early efforts, e.g. [19], are described) as it usually
requires some (combination of) typical ontology management activities (such as on-
tology evolution and knowledge sharing – please see some details above).
Each of the application categories sets up some requirements to specific alignment
methods used within the category. Due to the wide variety of ontologies used in
industry it is difficult to set up a detailed set of requirements for ontology matching
methods. These requirements may substantially vary depending on ontology size and
structure so we outline only the most general observations. We analyze the require-
ments for ontology alignment regardless to industrial application in [8].
Run-time. 1st and 2nd categories assumed the matching process to be performed
at run-time. In that case the maximum level of automation must be reached. In 3rd
102 M. Davidovsky, V. Ermolayev and V. Tolok
and 4th categories it is allowed to perform matching and relative activities previously
and separately. This allows active involvement of experts to the matching process (for
alignment validation, relevance verification, etc.).
Completeness. Completeness is of the most importance in the 1st and 3rd cases. It
is important not to miss knowledge in these cases. At the same time, in the 2nd
category the response time of method implementation to a system query is more
critical as in that case matching is usually performed during runtime.
Relevance. In the 4th case, the relevance of knowledge is the most critical
(particularly during migration from an older system to a newer one). Here it is first of
all important to save actual knowledge, but some obsolete knowledge may be
discarded.
3 Solution Overview
The main focus of the paper is evaluation of ontology alignment and instance migra-
tion methodology in industrial settings. The methodology assumes (semi-) automated
iterative process of ontology alignment and instance migration with possible human
intervention for checking the correctness and setting up the process. The overall
methodology consists of two steps: (1) mapping discovery and determination of struc-
tural differences between ontologies and (2) ontology instance transformation and
migration according to the determined differences.
The first step is essentially the process of ontology matching with the only differ-
ence that it results not only in ontology alignment but also produces an output of a set
of transformation rules that further drive the process of ontology instance migration.
The solution for the first step is based on the implementation of meaning negotiation
between intelligent agents (we call this agent-based solution ABOA matcher [8]). The
matching process embodies the strategy that originates from [20] and is described in
detail in [21]. Negotiations among the agents are conducted in an iterative way and
with an aim to reduce the semantic distance between the negotiated structural contexts
of the respective ontology schemas. A negotiation is stopped when the distance
reaches a commonly accepted threshold or the parties exhaust their propositions and
arguments.
At the second step agents use Instance Migration Engine in order to transfer in-
stances between ontologies based on the transformation rules generated at the first
step. Instance migration results in the transfer of all the assertions that do not require
the resolution of the problem cases by the ontology engineer. The cases that caused
problems are recorded in the migration log. The details on the second step of the
methodology are described in [22].
Application of an Instance Migration Solution … 103
4 Evaluation Experiment
To test our methodology and solution of ontology alignment and instance migration
we choose real industrial ontologies: freeClass1 ontology for construction and build-
ing materials and services and eClassOWL2 [14] – the web ontology for products and
services. The dataset of the European building and construction materials market for
the Semantic Web (BauDataWeb3) has been selected as the set of assertions for mi-
gration. Structural parameters of the ontologies are presented in Table 1. General
experimental set-up specified in ISO/IEC 24744 notation for describing methodolo-
gies [23] is pictured in Figure 1.
Table 1. Structural parameters of industrial ontologies used in the second experiment
Number of
Total number Number of Number of object Number of data Number of
logical axi-
of axioms classes properties properties individuals
oms
freeClass 78414 9622 5231 168 3 1335
eClassOWL 360243 117090 60662 4900 2453 4766
Over 60
BauDataWeb - - - - - million
instances
Ontology 1 Ontology 2 Ontology Reference Instance Ontology 1 Ontology 2 Instance
TBox (OWL) TBox (OWL) Mappings Alignment Transforma- ABox (OWL) ABox (OWL) Migration Ontology 1 Ontology 2
(Alignment (Alignment tion Rules Log ABox (OWL) ABox (OWL)
Format) Format)
r r c r m r m r r m
r r c
Next iteration c
Generate Migrate Check and Migrate
Discover Compare
I Mappings II III Transformation IV Instances V Correct VI Problem Cases
and Analyze Rules
Instance
Agents Domain Ontology Migration Ontology Ontology
Expert Engineer Engine Engineer Editor
Refine and
Supplement
Fig. 1. The set-up of the evaluation experiment
The test case doesn’t contain any reference alignment. Hence, we had to determine
reference mappings manually in order to objectively judge about the obtained results.
For convenience both freeClass and eClassOWL ontologies may be divided into 2
parts. The first parts are actually the sets of entities directly inherited from the
1
http://www.freeclass.eu/ – the ontology for construction and building materials and services
2
http://www.heppnetz.de/projects/eclassowl/ – the web ontology for products and services
3
http://semantic.eurobau.com/ – BauDataWeb: the European Building and Construction Mate-
rials Database for the Semantic Web
104 M. Davidovsky, V. Ermolayev and V. Tolok
GoodRelations ontology4 [13] and also some concepts from other common-sense
vocabularies. The schemas of those parts of the ontologies are almost identical, so the
difference is mostly in the sets of individual assertions. Further, those parts do not
cause any problems in the discovery of the reference mappings as the entities mainly
have human-understandable names and labels. Based on the analysis of the above-
mentioned parts of the ontologies we constructed the set of reference mappings (fur-
ther mentioned as Alignment 1). The second parts of the ontologies consist of internal
entities that do not have understandable names (the names represent some identifiers
composed of numbers and characters), but some of them still have labels with de-
scriptions. Due to the big quantity of those entities we did not analyze the whole sets
and choose the 20 entities that are semantically similar. Then we discovered respec-
tive mappings for those chosen entities (further mentioned as Alignment 2). The pa-
rameters for both alignments are presented in Table 2 where for brevity we include
only the information about the classes and properties.
Table 2. Parameters of reference alignment for the experiment with the BauDataWeb dataset
Number of mapped entities
Classes Object properties Datatype Properties
Alignment 1 53 55 53
Alignment 2 20 11 0
Thus, the experiment with the BauDataWeb dataset has been performed in two
phases. Within the first phase we constructed the reference alignment (Alignment 1)
and started the matching process using the ABOA matcher. Then we found the map-
pings that correspond to Alignment 1 and compared them to the reference ones.
Alignment quality values for the results of this step are very high (Table 3, row 1) as
these parts are almost identical.
Table 3. Matching results
Experiment Alignment Quality Measures
Step Precision Recall F-Measure
1 0.99999 0.99999 0.99999
2 0.69552 0.42384 0.52671
It might be considered that the Alignment 1 in our experiment is not a topically in-
teresting case as the semantic differences are tiny and could be easily discovered
manually. However, this experimental phase represents a good case for validating the
generated instance transformation rules and instance migration quality. In this phase
all of the generated transformation rules were correct. More details on the transforma-
tion rules could be seen in [22]. Within the second phase we determined the Align-
ment 2 and tried to find the respective mappings within the alignment discovered by
the matcher. The alignment quality measures for the second phase are lower than for
4
http://www.heppnetz.de/projects/goodrelations/ – the web vocabulary for e-commerce
Application of an Instance Migration Solution … 105
the phase 1, which is conditioned by the relatively weak semantic similarity between
the structural contexts [20] that correspond to these parts of ontologies. It is also
worth noticing that the Precision value is noticeably higher than the Recall one within
phase 2. It is so because string-based structural similarity measurement methods yield
high values on labels. Labels can contain parts (e.g. words) that are common for many
of them, but respective entities in general are not semantically similar. For example
the comparison of labels “construction technology” and “pump technology” will give
noticeably high similarity values. However those labels belong to the entities that are
obviously not that similar semantically.
5 Concluding Remarks and Future Work
The paper presented the experiment evaluating our methodology and software solu-
tion for ontology instance migration on real-world industrial ontologies. The experi-
ment shows acceptable results that allow a positive judgement about the applicability
of our methodology in industrial settings. The results also suggest some directions for
the future work. The experiment with large ontologies (BauDataWeb dataset) shows
that the ontology instance migration engine allows migrating about several million
instances using a conventional desktop computer. Hence, a technique to overcome
this upper limit is needed for scaling the tool up to the volumes characteristic to Big
Semantic Data. Looking for such a technique is on our research and development
agenda. In the future we also plan to conduct a series of experiments with the ontolo-
gies specified in OWL sublanguages5 and OWL 2 profiles6. Another important direc-
tion for the future research is evaluating our approach on ontologies having different
structural patterns like a taxonomy (tree-type) structure, a network structure (ontolo-
gies rich with object properties), OWL graphs with high and low vertex degrees, etc.
Acknowledgments
The authors are grateful to the colleagues who provided the industrial ontologies for
our experiments – Univ. Prof. Dr. Martin Hepp and Dipl. Ing. Andreas Radinger from
E-Business and Web Science Research Group of the Universität der Bundeswehr,
München.
References
1. Bittner, T., Donnelly, M., Winter, S.: Ontology and Semantic Interoperability. In: D. Pros-
peri and S. Zlatanova (ed.) Large-scale 3D Data Integration: Problems and Challenges.
London, CRCPress (2005)
5
http://www.w3.org/TR/2004/REC-owl-features-20040210/#s1.3
6
http://www.w3.org/TR/owl2-profiles/
106 M. Davidovsky, V. Ermolayev and V. Tolok
2. Malzahn, D.: Industrial Application of Ontologies. In: eKNOW 2011, The Third Interna-
tional Conference on Information, Process, and Knowledge Management, (2011)
3. Tatarintseva, O., Ermolayev, V., Fensel, A.: Is Your Ontology a Burden or a Gem? – To-
wards Xtreme Ontology Engineering. In: Ermolayev, V. et al. (eds.) Proc. ICTERI 2011,
CEUR-WS.org/Vol-716, 65-81 (2011)
4. Euzenat J., Shvaiko P.: Ontology Matching. Berlin Heidelberg, Springer-Verlag (2007)
5. Corcho, O.: Methodologies, tools and languages for building ontologies. Where is their
meeting point? Data & Knowledge Engineering 46, 41–64 (2003)
6. Ehrig, M.: Ontology Alignment: Bridging the Semantic Gap (Semantic Web and Beyond).
Springer (2006)
7. Zhdanova A. V., de Bruijn, J., Zimmermann, K., Scharffe, F.: Ontology Alignment Solu-
tion. Deliverable D14 v2.0, (2004)
8. Ermolayev, V., Davidovsky, M.: Agent-Based Ontology Alignment: Basics, Applications,
Theoretical Foundations, and Demonstration. Tutorial Paper. In: Dan Burdescu, D., Aker-
kar, R., Badica, C. (eds.) Proc. WIMS 2012, 11-22, ACM (2012)
9. Silver, G., Hassan, O.H., Miller, J.: From domain ontologies to modeling ontologies to ex-
ecutable simulation models. In: Proc. of the 2007 Winter Simulation Conference, (2007)
10. Novák, P, Šindelář, R.: Applications of ontologies for assembling simulation models of in-
dustrial systems. In: Proc. of the 2011th Confederated international conference on the
move to meaningful internet systems (OTM'11), pp.148–157, Springer-Verlag Berlin, Hei-
delberg (2011)
11. Ermolayev, V., Keberle, N., Matzke, W.-E.: An Upper Level Ontological Model for Engi-
neering Design Performance Domain, LNBIP, vol. 20, pp.127–141. Springer, Heidelberg
(2008)
12. Ding, Y., Fensel, D., Klein, M., Omelayenko, B., Schulten, E.: The Role of Ontologies in
eCommerce. In: Steffen Staab, Rudi Studer (eds.): Handbook on Ontologies. International
Handbooks on Information Systems. pp. 593-616, ISBN 3-540-40834-7, Springer (2004)
13. Hepp, M. GoodRelations: An Ontology for Describing Products and Services Offers on the
Web, LNCS, vol. 5268, pp. 332–347. Springer Berlin Heidelberg (2008)
14. Hepp, M.: Products and Services Ontologies: A Methodology for Deriving OWL Ontolo-
gies from Industrial Categorization Standards. In: Int'l Journal on Semantic Web & Infor-
mation Systems 2(1) (2006), pp. 72–99, (2006)
15. Morgenstern, L., Riecken, D.: SNAP: An Action-Based Ontology for E-commerce Rea-
soning. In: Proc., Formal Ontologies Meet Industry, Verona, Italy, (2005)
16. Berners-Lee, T., Hendler, J., Lassila, O.: The Semantic Web (Berners-Lee et. al 2001).
Scientific American 284, 28–37 (2001)
17. Izza, S., Vincent, L., Burlat, P.: A Unified Framework for Enterprise Integration: An On-
tology-Driven Service-Oriented Approach. In: Pre-proc. of the First International Confer-
ence on Interoperability of Enterprise Software and Applications (INTEROP-ESA’2005),
pp. 78–89. Geneva, Switzerland, February 23 – 25, (2005)
18. Stoutenburg, S. et al.: Ontologies in OWL for Rapid Enterprise Integration. Time 122, 82–
89 (1994)
19. Lochovsky, F. H., Woo, C. C., Williams, L. J.: A micro-organizational model for support-
ing knowledge migration. In: Proc. of the ACM SIGOIS and IEEE CS TC-OA conference
on Office information systems, pp.194–204. Cambridge, Massachusetts, US, (1990)
20. Ermolayev V., Keberle, N., Matzke, W.-E., Vladimirov, V.: A Strategy for Automated
Meaning Negotiation in Distributed Information Retrieval. In: Y. Gil et al. (Eds.): ISWC
2005 Proc. 4th Int. Semantic Web Conference (ISWC'05), 6–10 November, Galway, Ire-
land. LNCS 3729, pp. 201–215 (2005)
Application of an Instance Migration Solution … 107
21. Davidovsky, M., Ermolayev, V., Tolok, V.: Agent-based implementation for the discovery
of structural difference in OWL DL ontologies. In: Mayr, H. C., Ginige, A., Liddle, S.
(ed.) Proc. Fourth Int United Information Systems Conference (UNISCON 2012). LNBIP
137, Berlin, Heidelberg: Springer-Verlag (2013)
22. Davidovsky, M., Ermolayev, V., Tolok, V.: Instance Migration between Ontologies having
Structural Differences. Int. J. on Art. Int. Tools. 20(6), 1127-1156 (2011)
23. Henderson-Sellers, B., Gonzalez-Perez, C.: Standardizing Methodology Metamodelling
and Notation: An ISO Exemplar. In: Kaschek, R., Kop, C., Steinberger, C., Fliedl, G.
(eds.) UNISCON 2008. LNBIP, vol. 5, pp. 1–12. Springer, Berlin/Heidelberg, (2008)
Extracting Knowledge Tokens from Text Streams
Eugene Alferov1,2 and Vadim Ermolayev1
1
Department of IT, Zaporozhye National University,
66 Zhukovskogo st., 69063, Zaporozhye, Ukraine
alferov.evgeniy@gmail.com, vadim@ermolayev.com
2
Kherson State University, 27, 40 Rokiv Zhovnya ave., 73000, Ukraine
alferov_jk@ksu.ks.ua
Abstract. This problem analysis paper presents our position on how could the
solution be sought to the problem of extracting semantically rich fragments
from a stream of plain text posts. We first present our understanding of the
problem context and explain the focus of our research. Further, in the problem
setting section we elaborate the workflow for knowledge extraction from in-
coming information tokens. This workflow is then used as a key to structure our
review of the literature on the relevant component techniques which may be ex-
ploited in a combination to achieve the desired outcome. We finally outline our
plan for conducting the experiments with an aim to validate the workflow and
find a proper combination of the component techniques for all steps which may
solve our specific research problem.
Keywords. Workflow, knowledge extraction, text streams, processing, ontol-
ogy learning, component techniques
Key terms. Data, Process, Knowledge, Approach, Methodology
1 Introduction
The dramatic growth of data volumes we face today is accelerated by the increase of
social networking applications that allow non-specialist users create a huge amount of
content easily and freely. Equipped with rapidly evolving mobile devices, a user is
becoming a nomadic gateway boosting the generation of additional real-time sensor
data. The emerging Internet of Things makes each and every thing a data or content,
adding billions of additional artificial and autonomic sources of data to the overall
landscape. Smart spaces, where people, devices, and their infrastructures are all
loosely connected, also generate data of unprecedented volumes and with velocities
rarely observed before. Noticeably, the major part of the new data comes in streams.
An expectation is that valuable information will be extracted out of all these data to
help improve the quality of life and making our world a better place – for humans.
Extracting Knowledge Tokens from Text Streams 109
Humans are however left bewildered about how to use, analyze, understand all these
data, giving a proper account to its dynamics. A topical recent estimate of the need for
data-savvy managers in the United States is 1.5 million [1]. This manpower is needed
to extract and use valuable information and knowledge for further decision making.
The critical steps in this work are (i) extracting information and knowledge; and (ii)
bringing the descriptions of the reflections of the world or domain into a refined state
– accounting for the changes brought in by new data, at scale.
In this paper we focus on the step (i) extraction. In Section 2 we present the prob-
lem statement by giving basic definitions and providing our view on how could a
processing workflow look like. The plethora of approaches, techniques, technologies,
and software tools already exist for solving different parts of the overall problem.
Hence we analyze the related work and structure this analysis using the workflow as
the key in Section 3. Finally we conclude the paper and present our plans for the fu-
ture proof of concept experimental work in Section 4.
2 Problem Statement
Ontology is a complex artifact that comprises structural components of several types.
Further the structural denotation of an ontology used in Description Logics [2] is
exploited: an ontology O comprises its schema S and the set of individuals
I : O ( S , I ) . Ontology schema is also referred to as a terminological component
(TBox). It contains the statements describing the concepts of O, the properties of
those concepts, and the axioms over the schema constituents.
If a finer grained look at an ontology schema is taken, one may consider S
comprising the following interrelated constituents: S {S C , S O , S D , S A }, where S C
is the set of statements describing concepts, S O is the set of statements describing
object properties, S D is the set of statements describing datatype properties, and S A
is the set of axioms specifying constraints over S C , S O , and S D (c.f. [3]). One may
notice that these constituents correspond to the types of the schema specification
statements of an ontology representation language L which is used for specifying O .
The set of individuals, also referred to as assertional component (ABox), is the set
of the ground statements about the individuals and their attribution to the constituents
of the schema.
Ontology Learning is the process of extracting the abovementioned constituents of
O from a text stream source. More specifically, the problem which is approached in
this research work is twofold:
For every individual plain text document (further referred to as information token)
arriving in the stream window DO:
(i) Extract ontological fragment (further referred to as knowledge token) specifying
the semantics of the information token.
(ii) Refine the ontology O incorporating the changes brought in by the knowledge
token.
110 E. Alferov and V. Ermolayev
The focus of this paper is the first part of the problem – the extraction of knowl-
edge tokens from information tokens of plain text in a particular professional domain
coming in a stream. The texts of ICTERI paper abstracts have been chosen as the
domain and source text corpus for our initial experiments – see also Section 4.
As an ontology is a complex artifact, the extraction of knowledge tokens from texts
is also a complex process. It comprises several steps and, possibly, iterations for ex-
tracting different structural constituents of S and I . These steps produce several
types of outputs in a particular sequence, sometimes referred to as the ontology learn-
ing layer cake (c.f. [4]). Those outputs are terms – concepts and their instances –
datatype properties – taxonomic relationships and object properties – axioms. Based
on [5] we present in Fig. 1 a workflow putting together extraction steps, inputs, out-
puts, and required component technology types.
The overall workflow contains two consecutive phases – Text Pre-processing and
Ontology Extraction. Text Pre-processing phase gets the information token as a plain
text input and produces its structured representation as a set of terms by applying
several statistical and linguistic techniques. All the tasks of the Ontology Extraction
Phase use the output of Phase 1 as their input and incrementally build up the knowl-
edge token by adding different ABox and TBox constituents. For that statistical, lin-
guistic, semantic, and logical techniques are employed in combinations. Fig. 1 lists all
relevant component techniques per task. All of those are never used in implementa-
tions. Therefore our initial research objective is to find out which combination of
component techniques works best of all for our specific data – i.e. copes well with (a)
the texts of small size but belonging to a particular domain; and (b) limited processing
time constrained by a stream window lifetime parameter. Further, after this constella-
tion of component techniques is chosen, the objective would be to refine those which
do not provide results of a satisfactory quality in our problem settings.
3 Related Research and Available Component Techiques
In this section we will describe the component techniques, outlined in Fig. 1, which
we found relevant to our work. Those component techniques could overall be catego-
rized as linguistic, statistic, semantic and logical (c.f. [5]). As pictured in Fig. 1 they
could be applied at different steps and for different purposes. Though not explicitly
shown in Fig. 1, the steps may undergo iterations for refining their results. Therefore,
the workflow proposed in this paper could be considered as hybrid and iterative.
De-noising (statistical, linguistic). This is a method that extracts the de-noised text,
comprising the content-rich sentences, from full texts [6]. Processing of noisy text
becomes important because the quality of texts in the form of blogs, emails and chat
logs can be extremely poor. The sentences in dirty texts are typically full of spelling
errors, ad-hoc abbreviations and improper casing [7].
Tokenization. Tokenization is splitting the text into a set of tokens, usually words.
This process is unsupervised and can be performed automatically by progam-parser.
Part of speech detection/tagging (linguistic). Part of speech tagging (POST) is the
process of assigning one of the parts of speech to the given word. POST provides the
Extracting Knowledge Tokens from Text Streams 111
syntactic structures and dependency information required for further linguistic analy-
sis in order to uncover terms and relations. POST is a semi-supervised or even unsu-
pervised process.
PHASE 1 (T1): PHASE 2 (T2 – T6):
Text Pre-processing Ontology Extraction
Bag
of Terms
Information
Token
Form Concepts and Extract Datatype
T1 Extract Terms
- De-noising
T2
Concept Instances
T3
Properties
*
- Sentence parsing Domain - Co-occurrence analysis - Syntactic structure analysis
- Part of speech - Clustering - Dependency analysis
detection Concepts - Latent semantic analysis - Association rule mining
- Syntactic structure - Sub-categorization - Use of Lexico-syntactic
analysis frames patterns
Domain - Relevance analysis Semantic - Use of a Semantic lexicon Domain - Use of Semantic templates
Terms - Co-occurrence analysis Properties
- Logical inference
Lexicon
Knowledge Token
Bag
of Terms
Extract / Discover Extract Object
LEGEND: * T4 Concept Hierarchies T5 Properties T6 Extract Axioms
Control flow - Clustering - Association rule minng - Use of Axiom templates
Information flow - Term subsumption - Syntactic structure - Inductive logic
Information flow in case - Use of a Semantic lexicon analysis programming
- Syntactic structure - Dependency analysis
of (semi-) supervised approach analysis - Use of Semantic
Ontology / Resource - Dependency analysis templates
Workflow step (task) Semantic - Use of Semantic templates - Use of Lexico-syntactic
Lexicon - Use of Lexico-syntactic patterns
patterns - Logical inference
- Logical inference
Fig. 1. A workflow for knowledge token extraction
Lemmatization (linguistic). Lemmatization is the reduction of morphological
variants of the tokens to their base form that can be performed in unsupervised way.
For achieving this word form must be known, i.e. the part of speech of every word has
to be assigned in the text document. This process usually takes a time and may con-
tain errors.
Chunking (linguistic). Chunking is unsupervised splitting a text in syntactically
correlated parts.
Sentence parsing. Sentence parsing is identifying the syntactic structure of a sen-
tence, for example in a form of a parse tree.
Syntactic structure analysis (linguistic). In syntactic structure analysis, words and
modifiers in syntactic structures (e.g., noun phrases, verb phrases, and prepositional
phrases) are analyzed to discover potential terms and relations. It can be done in un-
supervised way.
112 E. Alferov and V. Ermolayev
Relevance Analysis (statisitcal). The extent of occurrence of terms in individual
documents and in text corpora is employed for relevance analysis. This is semi-
supervised or even unsupervised technique.
Co-occurrence analysis (statisitcal). Co-occurrence analysis identifies lexical
units that tend to occur together for purposes ranging from extracting related terms to
discovering implicit relations between concepts [5]. This technique is unsupervised.
Clustering (statistical). Grouping together variants of terms to form concepts and
separating unrelated ones is known as terms clustering. It usually unsupervised tech-
nique. In this approach some measure of similarity is employed to assign terms into
groups for discovering concepts or constructing hierarchy [8]. Some of the major
issues in clustering are working with high-dimensional data and feature extraction and
preparation for similarity measurement. This gave rise to a class of featureless simi-
larity measures based solely on the co-occurrence of words in large text corpora. It is
known that clustering results are of acceptable quality only if a statistically represen-
tative (i.e. large) text corpora is processed. This fact limits the applicability of this
technique in our settings (texts of small size). However, used in the combination with
other techniques, clustering may yield some valuable addition to the result – and thus
needs to be tried.
Latent semantic analysis (statistical). Latent semantic analysis (LSA) is a theo-
retical approach and mathematical method for determining the meaning similarity of
words and passages by analysis of large text corpora. The main idea is that the aggre-
gate of all the word contexts in which a given word does and does not appear provides
a set of mutual constraints that largely determines the similarity of meaning of words
and sets of words to each other [9]. LSA can be useful in our investigation because it
is a fully automatic mathematical and statistical technique for extracting and inferring
meaningful relations from the contextual usage of words in text.
Sub-categorization (linguistic, semantic). Sub-categorization, or extracting sub-
categorization frames, is an approach to extract one type of lexical information with
particular importance for Natural Language Processing (NLP). Access to an accurate
and comprehensive sub-categorization lexicon is vital for the development of success-
ful parsing technology important for many NLP tasks (e.g. automatic verb classifica-
tion) and useful for any application which can benefit from information about predi-
cate-argument structure (e.g. Information Extraction) [10].
Using semantic lexicon (linguistic, semantic). A semantic lexicon is a dictionary
or thesaurus of words/terms labeled with semantic classes (e.g., “ongoing effort” is an
Activity) so associations can be drawn between words that have not previously been
encountered [11]. Semantic lexicons are a popular resource in ontology learning and
play an important role in many NLP tasks.
Dependency analysis (linguistic). Syntactic structure consists of lexical items,
linked by dependencies. They are binary asymmetric relations that are held between a
head and its dependents. Dependency analysis examines dependency information to
uncover relations at the sentence level. In this analysis, grammatical relations, such as
subject, object, adjunct, and complement, are used for determining more complex
relations. Dependency analysis is usually unsupervised approach.
Extracting Knowledge Tokens from Text Streams 113
Association rule mining (statistical). Association rule mining aims to extract cor-
relations, frequent patterns, associations or casual structures among sets of items in
data repositories [12]. It is an unsupervised component technique which works well
for considerably big data corpora. Association rules highlight correlations between
features in the texts, e.g. keywords. Association rules can be easy interpreted and are
understandable for an analyst or even for a normal user.
Use of lexico-syntactic patterns (linguistic). Lexico-syntactic patterns (LSPs) are
generalized linguistic structures or schemas that indicate semantic relationships
among terms and can be applied to the identification of formalized concepts and
conceptual relations in natural language text [13]. Lexico-syntactic patterns are
suitable for automatic ontology building, since they model semantic relations. These
display exactly the kind of relation between their parts that makes them easily
translatable into an ontology representation.
Use of semantic templates (semantic, linguistic). Semantic templates are similar
to lexico-syntactic patterns in terms of their purpose. However, semantic templates
offer more detailed rules and conditions for extracting not only taxonomic relations
but also complex non-taxonomic relations [5].
Logical inference (logical, semantic). In logical inference implicit relations are de-
rived from existing ones using rules such as transitivity and inheritance [5]. However,
the introduction of invalid or conflicting relations may also happen in case of an in-
complete or underspecified inference rule set – for example because of improper ac-
count for the validity of transitivity or mutual disjointness axioms.
Term subsumption (statistical, semantic). In the subsumption method, a given
term subsumes another term if the documents in which the latter term occurs are a
subset of the documents in which the given term occurs [14]. A term subsumption
measure is used to quantify the extent of a term x being more general than another
term y. This technique is semi-supervised and unsupervised too. The term subsump-
tion technique is easy to implement and it makes labeling concepts an easy task.
However, with this method, it is difficult to classify terms that do not co-occur fre-
quently and it requires a large data set to work reliably.
Use of axiom templates (semantic, linguistic). Axioms are useful for describing
the relationships between the concepts of an ontology. They can be written in differ-
ent ways depending on the relation that exist among the concepts.
Inductive logic programming (logical, semantic). Inductive logic programming
(ILP) is a research area at the intersection of inductive machine learning and logic
programming. ILP generalizes the inductive and the deductive approaches by aiming
to develop theories, techniques and applications of inductive learning from observa-
tions and background knowledge represented in first order logical framework.
The overview of the applicability of the presented component techniques and their
interrelationship with respect to the tasks in our workflow are presented in Table 1.
114 E. Alferov and V. Ermolayev
4 Summary and Future Work
Our literature search has revealed that extracting knowledge, or more specifically
learning ontologies, from plain text corpora is a well developed research field that
continues to produce new results. However, and to the best of our knowledge, extract-
ing ontologies from text streams, with a constraint on the life time of an input infor-
mation token, is a recently emerged research problem. The reasons for adding this
specific problem to the research agenda are the phenomenon of Big Data, in particular
its velocity dimension, as well as the need for better, more reliable, semantically rich
solutions for automating Big Data analytics. One more complication introduced by
our problem setting is the small size of an individual information token which hinders
yielding good quality results using the majority of traditional statistical and linguistic
techniques for ontology extraction from text corpora.
We argued in this paper that applying a combination of the relevant existing com-
ponent techniques in a structured and iterative way may overall produce such a result
– as an incremental collection of ontology elements in a knowledge token provided by
individual techniques at different stages in our proposed workflow.
Table 1. Relevance of component techniques to the tasks within the workflow for extracting
knowledge tokens from information tokens
Task (Fig. 1.)
Component technology
T1 T2 T3 T4 T5 T6
De-noising st, li
Part of speech detection/tagging li
Lemmatization li
Chunking li
Syntactic structure analysis li li li li
Relevance Analysis st
Co-occurrence analysis st st
Clustering st st
Latent semantic analysis st
Sub-categorization se, li
Using semantic lexicon se, li se, li se, li
Dependency analysis li li li
Association rule mining st st
Use of lexico-syntactic patterns li li li
Use of semantic templates se, li se, li
Logical inference lo, se lo, se lo, se
Term subsumption st, se
Use of axiom templates se, li
Inductive logic programming lo, se
Legend : li – linguistic; lo – logical; se – semantic; st – statistical;
As this research is in an early phase, we do not yet have the proof for this hypothe-
sis. However there is the plan in place for conducting the initial series of the “proof-
of-concept” experiments in which the component technologies will be exploited in a
semi-supervised or supervised fashion. For that we plan to use a small but well se-
Extracting Knowledge Tokens from Text Streams 115
mantically annotated corpus of the abstracts (information tokens) and full texts of
ICTERI papers collected in the ICTERIWiki portal1. This document corpus is incre-
mentally extended by adding the papers and their semantic annotations for each new
ICTERI conference instance. The annotations are done using the ICTERI Scope On-
tology by Tatarintseva et.al. [15]. These annotations will be used as a “Golden Stan-
dard” for evaluating the results of automated knowledge token extraction using the
workflow proposed in this paper.
After the concept is proven and the constellation of the component techniques is
circumscribed, we plan to test the approach on one of the professional news portals.
Further, it is planned to extend the proposed knowledge extraction procedure to sen-
sor stream data processing.
References
1. Manyika, J., Chui, M., Brown, B., Bughin, J., Dobbs, R., Roxburgh, C., Hung Byers, A.:
Big data: the Next Frontier for Innovation, Competition, and Productivity. McKinsey
Global Institute (2011), http://www.mckinsey.com/insights/mgi/research/technology_and_
innovation/big_data_the_next_frontier_for_innovation
2. Nardi, D., Brachman, R.J.: An Introduction to Description Logics. In: Baader, F.,
Calvanese, D., McGuinness, D. L., Nardi, D., Patel-Schneider, P. F. (eds.) The Description
Logic Handbook, Cambridge University Press New York, NY, USA (2007)
3. Davidovsky, M., Ermolayev, V., Tolok V.: Instance Migration between Ontologies Having
Structural Differences. In: Int. J. on Artificial Intelligence Tools, vol. 20(6), pp. 1127–
1156 (2011)
4. Buitelaar, P., Cimiano, P., Magnini, B.: Ontology Learning from Text: an Overview. In:
Buitelaar, P., Cimmiano, P., Magnini, B. (eds.). Ontology Learning from Text: Methods,
Evaluation and Applications, IOS Press, Amsterdam (2005)
5. Wong, W., Liu, W., Bennamoun, M.: Ontology Learning from Text: a Look Back and into
the Future. ACM Comput. Surv., 44(4), Article 20, 36 pages.
http://doi.acm.org/10.1145/2333112.2333115 (2012)
6. Shams, R., Mercer, R. E.: Investigating Keyphrase Indexing with Text Denoising. In:
Proceedings of the 12th ACM/IEEE-CS Joint Conf. on Digital Libraries,
pp. 263–266, ACM (2012)
7. Wong, W., Liu, W., Bennamoun, M.: Enhanced Integrated Scoring for Cleaning Dirty
Texts. arXiv preprint arXiv:0810.0332. (2008)
8. Cimiano, P., Hotho, A., Staab, S.: Learning Concept Hierarchies from Text Corpora using
Formal Concept Analysis. Journal of Artificial Intelligence Research Archive, 24(1), 305–
339 (2005)
9. Landauer, T.K., Foltz, P.W., Laham, D.: Introduction to Latent Semantic Analysis.
Journal: Discourse Processes, 25(2-3), 259–284 (1998)
10. Preiss, J., Briscoe, T., Korhonen, A.: A System for Large-Scale Acquisition of Verbal,
Nominal and Adjectival Subcategorization Frames from Corpora. In: Annual Meeting.
Association for Computational Linguistics, 45(1), 912 (2007)
11. Thelen, M., Riloff, E.: A Bootstrapping Method for Learning Semantic Lexicons using
Extraction Pattern Contexts. In: Proc. ACL-02 Conf. on Empirical Methods in Natural
1
http://isrg.kit.znu.edu.ua/icteriwiki/
116 E. Alferov and V. Ermolayev
Language Processing, Association for Computational Linguistics, vol. 10, pp. 214–221
(2002)
12. Kotsiantis, S., Kanellopoulos, D.: Association Rules Mining: a Recent Overview. GESTS
International Transactions on Computer Science and Engineering, 32(1),
71–82 (2006)
13. Summary on Requirements on Lexico-Syntactic Patterns (Synthesis by PC),
http://www.w3.org/community/ontolex/wiki/Specification_of_Requirements/Lexico-
Syntactic_Patterns
14. De Knijff, J., Frasincar, F., Hogenboom, F.: Domain Taxonomy Learning from Text: the
Subsumption Method versus Hierarchical Clustering. Data & Knowledge Engineering,
(2012)
15. Tatarintseva, O., Borue, Yu., Ermolayev, V.: Validating OntoElect Methodology in
Refining ICTERI Scope Ontology. In: H.C. Mayr et al. (Eds.): UNISCON 2012, LNBIP
137, pp. 128--139 (2013)
1.3 Model-Based
Software System Development
Use of Neural Networks for Monitoring Beam
Spectrum of Industrial Electron Accelerators
Oleksandr Baiev, Valentine Lazurik and Ievgen Didenko
School of Computer Science, V. N. Karazin Kharkiv National University,
4, Svobody Sqr., 61022, Kharkiv, Ukraine
oleksandr.baiev@gmail.com, lazurik@hotmail.com,
ievgen.v.didenko@gmail.com
Abstract. This paper investigates technique for solving spectrometry
inverse problem the neural network as method for reconstruction of
electron beam spectrum using depth-charge curve. The inverse problem
turned into multivariable optimization and the form of spectrum is based
on proposed three-parameter model. Radial basis function network calcu-
lates the parameters of this model. We developed computational experi-
ment using Monte-Carlo technique to evaluate strengths and weaknesses
of proposed approach and compare neural networks with conventional
data evaluation methods.
Keywords. Neural nets, Inverse problems, Monte Carlo, Radiation tech-
nologies, Depth-charge curve
Key terms. ComputerSimulation, Methodology, MachineIntelligence
1 Introduction
One of the main characteristic of the irradiation processes is an energy of beam.
This parameter influences on absorbed dose in target. Therefore, standards for
radiation technologies [1, 2] predetermine the upper bound of beam energy to
prevent ionization of the object under irradiation. Because of accelerator fea-
tures, electrons in beam have different energy. Thus, the beam energy repre-
sented by some function, which shows relations between particles number and
their energies. This function called beam spectrum. In practice at least three
parameters define the spectrum: average (Eav ) and probably (Ep ) energies and
full width on half maximum (Ew ). In order to measure beam energy dosimet-
ric wedge and stack are widely used in centers of radiation technologies. These
devices allow to determine only average and probable energies of beam [1–6].
Of course, these two parameters does not allow to reconstruct full energy dis-
tribution. Thereby developing of new instruments and methods of dosimetric
measurements is actual problem.
Mentioned devices intend to measure distributions of absorbed dose or charge
[5, 6]. The measured depth-dose (depth-charge) curves relate to beam spectrum
Use of Neural Networks for Monitoring Beam Spectrum . . . 119
through Fredholm integral equation and finding exact spectrum is an ill-posed
inverse problem [7]. This means that evaluated spectrum obtained by conven-
tional mathematical methods can differ with true energy distribution. There
are, for example, method of least squares (MLS) or method of Tikhonov regu-
larization (MTR). Above all, important disadvantage of the MLS and MTR is
impossibility to include additional solution conditions, for example, correlations
between parameters, positivity and other. This lack can bring to violation of
conditions, given by physical lows. It should be mention that in common case
the neural networks (NN) solve approximation tasks and find solutions based
on existing precedents after supervised training [8–12]. So the one of the way
of improving dosimetry effectiveness is developing of methods for measurement
results evaluation based on neural networks. In order to apply NN for dosimetric
data processing it is necessary to solve next problems: select networks topology,
obtaining data for NN training, developing methods for data preprocessing and
interpretation, system for evaluation network effectiveness.
So current research is about feasibility of using neural networks for devel-
oping system of measurement results evaluation for beam spectrum monitoring
of industrial electron accelerators. We will discuss mathematical model of mea-
surement process, which was built in order to compile training set for network
learning procedure (Section 2). Section 3 describes methods under investigation.
In section 4, we will show approach for methods evaluation, which contains com-
putational experiment and comparison criteria. In section 5 given comparison
results of neural networks and conventional methods testing.
2 Physical process and mathematical model
In order to calculate radiation energy, it is a common practice in field of radiation
technologies to measure depth-dose curve by dosimetric wedge. However, the
works of recent years propose new devices based on measurement of depth-
charge curve that can realize on-line energy monitoring [3–6]. In this work, we
will consider mathematical abstraction of these devices and will build method
for beam spectrum controlling using depth-charge curve.
2.1 Devices
Device [5] consists of two plates only and intend to calculate probable energy
as a value which linearly depends on charge in first plate to sum charge ratio.
Measurer in [6] contains 10 absorbers. But in order to simplify average energy
calculation the plates were combined and authors use similar to [5] dependency.
Fig. 1 shows principal schema of measurer. Dosimetric stack consists of set
of plates - absorbers. The absorbers material is often aluminum, because of
radiation ruggedness. The electron beam falls on the sequence of plates. Electrons
stop at different depths depending on their energy. Thus, absorbers collect some
charge which can be measured by current integrators connected to corresponding
plate. The set of measured values represents the depth-charge curve.
120 O. Baiev, V. Lazurik and Ie. Didenko
f1 f2 f3 f4 f5 fn
Fig. 1. Common schema of stack for depth-charge measurement
Mathematical model of the measurement process is based on a semi-empirical
model of the depth-charge distribution for monoenergetic electrons and model
of charge measurement uncertainty. Direct problem describes relation between
known beam spectrum and depth-charge curve through equation:
ZER
f (x) = Q(x, E)y(E) dE, x ∈ [0, xR ], (1)
EL
where y(E) - describes relation between number of particles and their energy
(electrons spectrum), f (x) - describes depth distribution of charge, xR mea-
surer full width, [EL , ER ] - operating energy range of accelerator, integral kernel
Q(x, E) corresponds to radiation type (α, β, γ) and measurer internal char-
acteristics (including absorbers material). Works [13, 14] describe appropriate
relations for monoenergetic beam and depth-charge curve.
In the research we neglect charge leakage and suppose that distance between
absorbers is neglectfully small. It means that each particle from initial beam
can stops in absorbers and pass through current integrator or can pass through
whole device with no impact in depth-charge curve.
The measurement results of charge distribution in absorbers is set f =
{f1 , f2 , . . . , fn } (see Fig. 1), where n - number of absorbers, fi - integral of
f (x) over the depth for i-th absorber:
ER
xiZ+∆xZ
fi = Q(x, E)y(E) dE dx, (2)
xi EL
where ∆x - absorbers width. Equation (2) can be approximated as:
fi = ∆x
P E
2 pj yj [Q(xk + (i − 1)∆x, Ej )+
j (3)
+Q(xk + i∆x, Ej )],
where i = 1, n, j = 0, m, m = (ER − EL )/∆E - number of steps of function y(E)
discretization over energy axis, ∆E - step of spectrum energy discretization, yj
- value of y(E) in approximation nodes, coefficient pE
j defines method and step
Use of Neural Networks for Monitoring Beam Spectrum . . . 121
of function y(E) approximation. Then the measurement process can be shown
as system of linear equations:
Ay = f⇔
a1,1 a1,2 · · · a1,n y0 f1
a2,1 a2,2 · · · a2,n y1 f2
, (4)
. . = .
.. .. . .
. . . .. .. ..
am,1 am,2 · · · am,n ym fn
where elements of matrix A are:
ai,j = ∆x E
2 pj [Q(xk + (i − 1)∆x, Ej )+ (5)
+Q(xk + i∆x, Ej )].
In order to approximate y(E) by method of trapezoids, coefficients pE
j are:
∆E/2 j = 0 ∨ j = m
pE
j = . (6)
∆E otherwise
It’s obvious that complexity of spectrum reconstruction grows with increasing
of m (dimension of vector y). In order to reduce problem the we used parameter-
ization of y(E). As mentioned above, the general practice is denoting spectrum
by parameters: Ep , Eav , Ew . Therefore, it is reasonable to make model of the
beam spectrum, which use three parameters.
2.2 Model of electrons spectrum
Fig. 2 shows geometrical interpretation of electrons spectrum model considered
in the present work. The graph of spectrum consists of two part: left exponential
and right linear slopes. The parameters of this model are:
– Emax – maximal particles energy in the beam,
– Ep – most probable energy,
– Es – energy of 10 times decreasing of the intensity compared to Ep electrons
along left slope.
In the future discussion the Π will denotes set of spectrum parameters, i.e.
Π = {Es , Ep , Emax }.
Parameters of the model correspond to characteristics of beam used in prac-
tice according to:
Ep = Ep ,
E −E
Ew = ln0.5 − Es ) + max2 p
ln0.1 (Ep (7)
Emax −Ep 0.45(Es −Ep )
Eav = Es + ln 4 + ln0.1
and mathematical expression for spectrum is:
122 O. Baiev, V. Lazurik and Ie. Didenko
h
0.5h
0.1h
Es Ep Emax
Fig. 2. Model of electron beam spectrum
heµ(E−Ep ) , 0 < E ≤ Ep
y(E) = k1 E + k2 , Ep < E ≤ Emax , (8)
0, Emax < E
ln(0.1) h hEmax
µ= , k1 = , k2 = , (9)
Es − Ep Ep − Emax Emax − Ep
where E ∈ [0; ∞], h = y(Ep ) - maximum of function y(E) and was obtained
with supposition of
EZmax
y(E)dE = 1. (10)
Es
Therefore, maximum of energy distribution is:
Es − Ep
h = y(Ep ) = [0.9 + 0.5(Emax − Ep )]−1 . (11)
ln(0.1)
It should be mention, that in accordance to physical laws the function y(E)
is positive or equal zero for all accepted E and parameters correlates as:
0 < Es < Ep ≤ Emax . (12)
2.3 Model of measurement
In the real experiment measured fi differ with its real value. This error grounded
on weaknesses of measurer and external influence. We will mark set of true values
of f (x) as f , and use f˜ to mark set of values complemented with measurement
uncertainty:
f˜ = (1 + εξ)f, (13)
where ε - value of standard deviation of measurement error, ξ - random variable
distributed in accordance to standard normal distribution:
Use of Neural Networks for Monitoring Beam Spectrum . . . 123
p
ξ = cos(2πr1 ) −2ln(r2 ), (14)
where r1 , r2 - random variables which ate distributed in accordance with stan-
dard uniform distribution.
We will use similar signature to denote evaluated parameters Π̃, Ẽs , Ẽp , and
Ẽmax reconstructed spectrum ỹ instead their true values without tilde.
3 Methods for spectrum reconstruction
3.1 Neural networks
In order to apply NN for solving spectrometry inverse problem reconstruction of
spectrum can be represented as multivariable function fitting. Suppose that func-
tion φ implements measurement process of depth-charge curve, i.e. f˜ = φ(Π).
Therefore, inverse function Π̃ = φ−1 (f˜) realizes transformation from depth-
charge curve to beam spectrum. So approximation of φ−1 can be used to get
spectrum using depth-charge curve. In the work we used general regression neu-
ral network (GRNN) [15] to fit φ−1 . This network needs set of precedence for
supervised learning. Consider algorithm of training set creation.
Implemented measurement models allow to create pairs s = (f˜, Π), where f˜
calculates from parameters set Π. The collection of s is based on different Π and
represents a reference points for φ−1 fitting:
f˜1 f˜2 · · · f˜N
(s1 · · · sN ) = (15)
Π1 Π2 · · · ΠN
where N - number of elements in training set. For future discussion, we will
denote each unique Π in training and testing sets as reference spectrum. Note,
that each values of parameters for all Π from training set was normalized in
accordance to [EL ; ER ] → [0; 1]. Of course, outputs of network were scaled back
during testing.
3.2 Conventional methods
Consider methods, which is traditionally used for measurement results evalua-
tion. The data which were obtained by these methods is a base level to determine
NN effectiveness for solving spectrum reconstruction problem. The method of
least squares calculates parameters Π as:
Π̃M LS = arg min AỹΠ − f˜ , (16)
Π
where k·k - Euclidian norm. Method of Tikhonov regularization expands MLS
through additional stabilizer function:
Π̃M T R = arg min AỹΠ − f˜ + α kỹΠ k , (17)
Π
124 O. Baiev, V. Lazurik and Ie. Didenko
where α > 0 - regularization parameter. It should be remind that using of math-
ematical model of measurement process gives true values of electrons spectrum.
So α can be calculated from [7]:
ky − ỹα k
α∗ = arg min . (18)
α kyk
In the work we applied Nelder-Mid simplex method numerical solution of
(16), (17) and (18).
4 Algorithm for evaluation methods preparing and
testing
4.1 Comparison approach
Implemented models of spectrum, measurement process and methods for data
evaluation compose computational experiment (Fig. 3 shows sequential dia-
gram). The experiment aim is comparison of methods for spectrum reconstruc-
tion. The approach which was used to build experiment uses Monte-Carlo tech-
nique: system generate measurement results, each methods reconstruct spectra
using samples of depth-charge cure, system calculates statistical characteristics
of reconstruction error. Computational experiment consists of three steps: prepa-
ration, main part (loop Common) and results interpretation.
Preparation of an experiment includes setting parameters of models and
methods. Main part is a series of subexperiments with varied measurement un-
certainty ε. Each of them contains two steps: training of NN and selected meth-
ods comparison. Both processes include generation of pairs s = (f˜, Π) which is
based on predefined set of Π. But these sets of reference spectrum are different.
Testing procedure (loop Data Evaluation) repeats sampling of f˜, evaluates ap-
propriate Π by each method and collects reconstruction error based on truth and
calculated spectra based on proposed set of indicators. The results processing
step aims to build relationships that show correlations between accuracy of spec-
trum reconstruction and varied error of measurement.
Software for experiment execution implemented in MATLAB with Neural
Network Toolbox (function newgrnn as NN), Optimization Toolbox (function
fminsearch as MLS and MTR). In order to speed up computational experiment,
software was executed on high performance cluster [16] with Distributed Com-
puting Toolbox.
4.2 Comparison indicators
In order to assess the effectiveness of methods for reconstruction of beam energy
characteristics we suggested set of indicators. The set consists of the standard
statistical estimates of data evaluation error and indicator of methods reliability.
There are two indicators type: mismatch along energy axis (estimate shift of
reconstructed spectrum along horizontal axis) and common indicator. Consider
details of each indicators.
Use of Neural Networks for Monitoring Beam Spectrum . . . 125
Fig. 3. Sequential diagram of computer experiment
1. Mismatch along energy axis. Average M (r) and standard deviation σr of
distance along intensity axis between reconstructed and true spectra are
based on:
1
r = (y − ỹ)2 ; (19)
n
2. Common characteristics. Probability of method failure P . We suppose that
the method failure is a case when applying mathematical methods leads to
impossible (due to physical lows) solution, i.e. the solution brakes condition
(12). It is obvious that value 1 − P characterize method reliability.
5 Results and discussions
5.1 Parameters of computation experiment
In order to evaluate methods effectiveness with suggested indicators we made
computational experiment with parameters shown in Table 1.
The training and testing sets include reference spectra with parameters:
126 O. Baiev, V. Lazurik and Ie. Didenko
Table 1. Common experiment parameters
Parameter Value
Characteristic for measurement depth-charge curve
Absorbers material Aluminum (Z = 13, Am = 27)
Absorber’s width (∆x) 0.4 g/cm2
Device total width (xR ) 6 g/cm2
Uncertainty (ε) Varied from 0% to 30%, step 1%
[EL ; ER ] [0M eV ; 10.2M eV ]
Ew of reference spectra Randomly from 2% to 10% of Ep
y(E) discretization step (∆E) 0.05M eV
Number of reference spectra 9000 (training) and 41000 (testing)
Ep = r1 , r1 ∼ U [EL , ER ],
Emax = Ep (1 + 2r2 ), r2 ∼ U [0.01, 0.02], (20)
ln0.1
Es = Ep − ln0.5 r3 , r3 ∼ U [0.01, 0.08].
Fig. 4 shows examples of sampled reference spectra. Number of the spectra
for training and testing sets is reduced, but proportion saved. As shown on Fig.
4 and in Table 1 the testing set is bigger than training set. It is necessary to get
appropriate assessment of method based on NN with influence of retraining.
Testing spectra Training spectra
1.8 2.5
1.6
1.4 2
1.2
y(E) − Intensity
y(E) − Intensity
1.5
1
0.8
1
0.6
0.4 0.5
0.2
0 0
5 6 7 8 9 10 5 6 7 8 9 10
E − Energy E − Energy
Fig. 4. Reference spectra for a) NN training and b) methods testing
As shown in Table 1 the device consists of 15 absorbers. This configuration
is chosen based on previous research [18] which was aimed to find optimal dis-
cretization step of depth-charge curve for spectrum reconstruction by NN. It
should be mention that in works [17, 18] sets for methods testing and prepara-
tion based on reference spectra with fixed Ew parameter and same maximum
h = 1. Therefore, seeking of optimal absorbers width is open for future research.
Use of Neural Networks for Monitoring Beam Spectrum . . . 127
M(r) σr Failures
0.25 0.5 0.5
MLS
0.2 0.4 0.4 MTR
GRNN
M(r) 0.15 0.3 0.3
σr
P
0.1 0.2 0.2
0.05 0.1 0.1
0 0 0
0 0.1 0.2 0.3 0 0.1 0.2 0.3 0 0.1 0.2 0.3
ε ε ε
Fig. 5. Results of methods comparison
5.2 Results and discussion
Fig. 5 contains obtained dependencies, which describe relation between methods
evaluation error and measurement uncertainty. The charts 5a and 5b based on
indicator (19). Chart 5c shows probability of method failure.
For MLS and MTR experiment proves expected results. Methods are sensi-
tive to uncertainty in input data. Fig. 5a and 5b show that error of MLS and
MTR solutions rapidly grows with increasing of ε. With respect to probability
of failure, both methods demonstrated almost equal inefficiency. It can be mean
that stabilizing additions in MTR does not affect to the method reliability. It
should be mention that the reason of MLS and MTR error for ε = 0% is dis-
cretization inaccuracy which appears when transforming integral (1) to system
(4).
As an opposite to conventional methods, the solutions obtained by NN have
smaller dependency between evaluation error and input data uncertainty. Fur-
thermore as shown on Fig. 5a, 5b the GRNN evaluates spectra more accurate
than MLS and MTR for measurement uncertainty more than 5-7%. The main
advantage of NN method is that GRNN reconstruct beam spectrum parame-
ters with no failures (see Fig 5c), i.e. all obtained solutions are compliance with
physical lows.
6 Conclusion
The work shows GRNN method effectiveness for solving inverse dosimetry prob-
lem of electron spectrum reconstruction using depth-charge curve. The main ad-
vantages of proposed technique compared to conventional methods is allowance
to apply additional solutions conditions. It lids to getting robust evaluation
method. As shown in the work methods based on NN can be used for building
on-line energy monitoring systems in centers of radiation technologies.
Furthermore, we proposed comparison approach based on Monte-Carlo tech-
nique and set of effectiveness indicators. The approach allows testing different
128 O. Baiev, V. Lazurik and Ie. Didenko
types of evaluation methods and can be used for methods optimization in order
to select or apply technique for industrial problems solving.
References
1. Standard ISO/ASTM 51649-2005(E). Practice for dosimetry in an electron beam
facility for radiation processing at energies between 300 keV and 25 MeV. United
States, 30 p. (2005)
2. ICRU Report 35. Electron beams with energies between 1 and 50 MeV. United
States, 160 p. (1984)
3. Fuochi P.G., Lavalle M., Martelli A., Corda U., Kovacs A., Hargittai P., Mehta
K., Electron energy device for process control, Radiation Physics and Chemistry,
Volume 67, pp. 593-598 (2003)
4. Fuochi P.G., Lavalle M., Martelli A., Corda U., Kovacs A., Hargittai P., Mehta K.,
Energy device for monitoring 4-10 MeV industrial electron accelerators, Nuclear
Instruments and Methods in Physics Research A, Volume 546, pp. 385-390 (2005)
5. M. Lavalle, P.G. Fuochi, A. Martelli, U. Corda, A. Kovacs, K. Mehta, and F.
Kuntz, Energy Monitoring Device for Electron Beam Facilities, International Top-
ical Meeting on Nuclear Research Applications and Utilization of Accelerators,
Conference proceedings, Vienna (2009)
6. Vanzha S.A., Nikiforov V.I., Pomatsalyuk R.I., Tenishev A.Eh., Uvarov V.L.,
Shevchenko V.A., Shlyakhov I.N., Development “radiation shadow” technique for
regime monitoring of product sterilization by electron beem, Problems of Atomic
Science & Technology. Series “Nuclear Physics Investigations”, Volume 2(53),
pp. 150–153 (2010)
7. Petrov Yu.P., Sizikov V. S., Well-Posed, Ill-Posed, and Intermediate Problems with
Applications, V.S.P. Intl Science, Leiden, Netherlands, 234 p. (2005)
8. Haykin S, Neural networks: a comprehensive foundation, 2nd edn. Prentice Hall,
Englewood Cliffs, United States, 936 p. (1999)
9. Michael M. Li, Brijesh Verma, Xiaolong Fan, Kevin Tickle, RBF neural networks
for solving the inverse problem of backscattering spectra, Neural Computing and
Applications, Volume 17, pp. 391-397 (2008)
10. Michael M. Li, William Guo, Brijesh Verma, Kevin Tickle, John OConnor, Intelli-
gent methods for solving inverse problems of backscattering spectra with noise: a
comparison between neural networks and simulated annealing, Neural Computing
and Applications, Volume 18, pp. 423-430 (2009)
11. Barradas N. P., Vieira A., Artificial neural network algorithm for analysis of
Rutherford backscattering data, Phys Rev E Stat Phys Plasmas Fluids Relat In-
terdiscip Topics 62, pp. 58185829 (2000)
12. Barradas N.P., Patricio R.N., Pinho H.F.R., Vieira A., A general artificial neu-
ral network for analysis of RBS data of any element with Z between 18 and 83
implanted into any lighter one- or two-element target, Nuclear Instruments and
Methods in Physics Research B, Volumes 219-220, pp. 105-109 (2004)
13. Adadurov A., Lazurik V., Rogov Yu., Tokarevskii V., Shechenko S., Spectrome-
try of Intense Fluxes of Gamma Radiation by Means of the Method of Capsule-
Absorbers, IEEE Nuclear Science Symposium and Medical Imaging Conference,
Conference Publications, Anaheim, Unated States, pp. 17 (1996)
14. Lazurik V.T., Lazurik V.M., Popov G., Rogov Yu., Zimek Z., Information System
and Software for Quality Control of Radiation Processing. IAEA: Collaborating
Use of Neural Networks for Monitoring Beam Spectrum . . . 129
Center for Radiation Processing and Industrial Dosimetry, Warsaw, Poland, 220 p.
(2011)
15. Specht Donald F., A General regression neural network, IEEE Transactions on
Neural Networks, Volume 2(6), pp. 568–576 (1991)
16. Baiev O., Didenko I., Lazurik V., Mishchenko V., Towards the questions on planing
the development of the department compute cluster, Proceedings of ICTERI-2011,
Kherson, Ukraine, pp. 27–28 (2011)
17. Baiev O., Lazurik V., Advantages of neural networks for deriving an electrons
spectrum from depth-charge curve, “IEEE Nuclear Science Symposium and Medi-
cal Imaging Conference”, Conference Publications, Valencia, Spain, pp. 1395-1397
(2011)
18. Baiev O.U., Lazurik V.T., Discretization grid of depth-charge curve selecting for
electrons beam spectrum reconstruction problem, Bulletin Kherson National Tech-
nical University, Value 3(42), pp. 62–66 (2011)
Lazy Parallel Synchronous Composition of
Infinite Transition Systems
Yuliia Romenska and Frédéric Mallet
Université Nice Sophia-Antipolis Aoste Team Project (INRIA/I3S),
Sophia Antipolis, France
Yuliia.Romenska@inria.fr, Frederic.Mallet@unice.fr
Abstract. Embedded System Design is becoming a field of choice for
Model-Driven Engineering techniques. On the engineering side, models
bring an abstraction of the code that can then be generated (and regen-
erated) at will. On the semantic side, they bring a reasoning framework
to guarantee or verify properties on the generated code. We focus here
on the Clock Constraint Specification Language, initially defined as a
companion language of the uml Profile for marte. More specifically,
we define a state-based representation of ccsl operators. To deal with
unbounded operators, we propose to use lazy evaluation to represent in-
tentionally infinite transition systems. We provide an algorithm to make
the synchronized product of such transition systems and we study its
complexity. Even though the transition systems are infinite, the result of
the composition may become finite, in which case the (semi)algorithm
terminates and exhaustive analysis becomes possible.
Keywords. Multiform logical time, synchronized product, lazy evalua-
tion, marte ccsl.
Key terms. FormalMethod, VerificationProcess, MathematicalModel,
SpecificationProcess.
1 Introduction. Context and Goal of the Project
In the model-driven approach to embedded system engineering, application and
architecture models are developed and refined concurrently, and then associated
by allocation relationships. The representation of requirements and constraints
in this context becomes itself an important issue, as they guide the search for
optimal solutions inside the range of possible allocations. One of the important
aspects of embedded system modeling is to capture the functional and non-
functional requirements as well as the constraints (functional and non-functional)
imposed by the execution platform through the allocation.
Multiform logical time is a flexible notion of time suitable for both functional
and extra-functional properties that supports an iterative refinement process.
Logical time considers time bases that can be generated from sequences of events
Lazy Parallel Synchronous Composition of Infinite Transition Systems 131
not necessarily regular in physical time (as the usual meaning suggest). Some
of the essence of multiform logical time was captured and encapsulated into a
dedicated language called the Clock Constraint Specification Language (ccsl)
[1, 2]. ccsl was initially defined as a companion language of the uml profile for
Modeling and Analysis of Real-Time and Embedded systems (marte) [3].
ccsl has arisen from different models in an attempt to abstract away the
data and the algorithm and to focus on events and control. Even though ccsl
was initially defined as the time model of the uml profile for marte, it has now
become a full fledged domain-specific modeling language for capturing chrono-
logical, causal and timed relationships and is now developed independently. It
combines constructs from the general net theory and from the synchronous lan-
guages [4]. It is based on the notion of clocks which is a general name to denote
a totally ordered sequence of event occurrences. It defines a set of clock relations
and expressions. Some ccsl operators are bounded, others are unbounded. The
bounded operators can be represented with finite Boolean transition systems.
Unbounded operators require a specific symbolic representation.
Until then, the clock calculus on ccsl specification was performed step by
step up to a predefined number of steps. This work is an attempt to support
exhaustive analysis of ccsl specification. When ccsl operators are represented
as transition systems, their composition is the synchronized product of the tran-
sition systems. However, this causes termination problems when the transition
systems have an infinite number of states. In this paper, an algorithm for the
parallel execution of automata representing ccsl operators is proposed. It has
been implemented in a prototype tool. This algorithm supports ccsl unbounded
operators. The infinite data structure is unfolded on demand using a lazy evalua-
tion technique. This is a significant evolution on previous verification techniques
for ccsl [5,6] that were only considering a subset of operators a priori bounded.
2 Contribution
The main contribution is to propose an encoding based on lazy evaluation to rep-
resent ccsl unbounded operators. The second contribution is to propose an algo-
rithm to build the synchronized product of such automata. The (semi)algorithm
terminates when the composition of unbounded automata becomes bounded.
In this work, the main operators of the clock constraint language were con-
sidered. For each basic expression and each relations of the kernel language, a
transition system is proposed. Each transition is labeled by the set of clocks that
must tick for the transition to be taken. Clocks can stall, tick or be dead. When
a clock is dead, it cannot tick anymore. A path in the automaton is then an
infinite word on the alphabet of powersets of clock names.
The automata representing the unbounded ccsl operators consist of an in-
finite number of states and therefore transitions (even though each state has
a finite number of outgoing transitions). For those operators, the lazy evalua-
tion technique was applied. It allows postponing the construction of the state
of an unbounded automaton to the moment when it is actually needed. In very
132 Yu. Romenska and F. Mallet
frequent cases, the specification becomes bounded and the intentional infinite
representation is never actually expanded.
On these transition systems, we apply the classical synchronized product
of transition systems [7]. In the worst case (when automata are independent)
the composition is very costly, exponential in the number of clocks. In some
(frequent) cases, the cost of composition is much better, even though we have
not identified yet the exact cases where the composition is tractable.
3 Related work and inspirations
3.1 Timed automata
The formalism of timed automata [8] has been designed to allow the specification
and verification of real-time systems. It extends the concept of finite ω-automata
by establishing time constraints. One of the main notions of the timed automata
theory is a clock (not to be confused with the notion of clocks introduced in
ccsl). In a timed transition table the selection of the next state depends on
an input symbol and the time reading of the symbol. For this purpose each
transition table is associated with the set of real-valued clocks. A clock can be
set to zero simultaneously with the execution of one of the transition. At any
instant the values of such clocks are equal to the time elapsed since the last time
they were reset. The transition can be executed only if the value of the clock
satisfies the constraint associated with this transition. Therefore in the theory
of the timed automata a clock is an entity intended for determination of time
which elapsed since the last execution of the transition and the setting the value
of the clock to zero.
To answer the question if this formalism can be applied for representation
of the ccsl language basic constructions, the definition of a clock in ccsl time
model must be considered. In terms of ccsl time model a clock is a set of
ordered instants. Each clock has a lifetime limited by birth and death instants.
Formally, a Clock c is a tuple (Ic , ≺c , c↑ , c↓ , ≡c↓ ) where Ic is a sequence of
instants (it can be infinite), c↑ , c↓ are birth and death instants respectively such
that Ic ∩ {c↑ , c↓ } = , ≡c↓ is a coincidence relation and ≺c is an order relation
on Ic ∪ {c ↑, c ↓}. All instants of clocks are strictly ordered, the birth instant
precedes all the other instants of the clock and every instant precedes the death.
If the set of instants Ic is infinite then the death instant is not necessary. Ic
represents the occurrences or ticks of the clock c.
Thus we can see that the notions of clock in the terms of timed automata
formal model and the clock constraint specification language are radically dif-
ferent. Timed Automata, clocks captured physical real-valued time properties,
whose value is within a dense time interval. All time clocks evolve at the same
rate (without drift). In ccsl, clocks represent logical time properties. The un-
bounded nature precisely comes from the relative (unbounded) drifts between
the clocks that evolve at their own independent rhythm.
Lazy Parallel Synchronous Composition of Infinite Transition Systems 133
3.2 Synchronous Data Flow, Marked Graphs
Synchronous Data Flow (SDF) [9, 10] is a special case of the dataflow model
of computation. Its main characteristic is that the flow of control is completely
predictable at compile time. The main components of SDF are the actors, to-
kens, and arcs. Production and consumption of tokens by actors allow modeling
of relative rates of events. In synchronous dataflow numbers of consumed and
produced tokens are constant throughout the execution. To avoid the overflow of
resources and to maintain a balanced system, the scheduler must fire the source
and destination components at different rates. Such systems can then be used
to capture the relative drift between ccsl clocks.
Safety analysis on SDF graphs is a way to determine whether the system
remains bounded or not. Such techniques could be used to study boundness
issues for ccsl specifications. However, this is not the concern of this paper. We
assume that the composition is bounded and propose an algorithm to build the
synchronized product.
3.3 Synchronized Product of Transition Systems
When ccsl operators are expressed as transition systems, their parallel com-
position simply is the synchronized product of the transition systems [7, 11, 12].
Synchronization vectors are used to decide which transition systems must syn-
chronize on which transitions. Synchronization vectors allows the specification
of purely asynchronous compositions (where only one single system is fired at
each step) to purely synchronous compositions (where all the automata must fire
one transition at each step), and all the intermediate synchronization schemes.
The main difference here is that the number of states may be infinite and we use
lazy evaluation to dynamically expand the states whenever they are required to
build a new composite state. The composition algorithm terminates only when
the synchronized product becomes finite. In [13], there was an initial attempt
to build the synchronized product of unbounded ccsl operators. In that work,
the automata were folded using extended automata (with unbounded integer
variables) rather than lazy evaluation. Therefore, the algorithm to compute the
synchronized product was always guaranteed to terminate. However, deciding
whether the result was finite or not would then require using integer linear pro-
gramming techniques.
4 The Clock Constraint Specification Language
This section briefly introduces the logical time model of the Clock Constraint
Specification Language (ccsl). A technical report [1] describes the syntax and
the semantics of a kernel set of ccsl constraints.
A clock c is a totally ordered set of instants, Ic . In the following, i and
j are instants.
S A time structure is a set of clocks C and a set of relations on
instants I = c∈C IC . ccsl considers two kinds of relations: causal and temporal
134 Yu. Romenska and F. Mallet
ones. The basic causal relation is causality/dependency, a binary relation on
I :4 ⊂ I × I. i 4 j means i causes j or j depends on i. 4 is a pre-order on I, i.e.,
it is reflexive and transitive. The basic temporal relations are precedence (≺),
coincidence (≡), and exclusion (#), three binary relations on I. For any pair of
instants (i, j) ∈ I × I in a time structure, i ≺ j means that the only acceptable
execution traces are those where i occurs strictly before j (i precedes j). ≺ is
transitive and asymmetric (reflexive and asymmetric). i ≡ j imposes instants i
and j to be coincident, i.e., they must occur at the same execution step, both of
them or none of them. ≡ is an equivalence relation, i.e., it is reflexive, symmetric
and transitive. i # j forbids the coincidence of the two instants, i.e., they cannot
occur at the same execution step. # is irreflexive and symmetric. A consistency
rule is enforced between causal and temporal relations. i 4 j can be refined
either as iπj or i ≡j, but j can never precede i. We consider here discrete sets of
instants only, so that the instants of a clock can be indexed by natural numbers.
For a clock c ∈ C, and for any k ∈ N>0 , c[k] denotes the k th instant of c.
Specifying a full time structure using only instant relations is not realistic
since clocks are usually infinite sets of instants. Thus, an enumerative spec-
ification of instant relations is forbidden. The Clock Constraint Specification
Language (ccsl) defines a set of time patterns between clocks that apply to
infinitely many instant relations.
4.1 The kernel relations
Table 1 gives a full list of the basic clock relations provided in the ccsl kernel.
For each of them the automaton was built. It is supposed that the automaton
can fire only if one of the participant clocks in ccsl operator ticks.
Table 1. Basic relations defined in the ccsl kernel (a and b are clocks, not instants).
Ref Name Kind of relation Notation
R1 Subclocking Synchronous a ⊂ b
R2 Coincidence Synchronous a = b
R3 Precedence Asynchronous, unbounded a 4 b
R4 Strict Precedence Asynchronous, unbounded a ≺ b
R5 Exclusion Asynchronous a # b
Coincidence According to the considered relation the clocks a and b always
tick simultaneously (a = b), it is defined as (∀k ∈ N? )(a[k] ≡ b[k]).
Subclocking a ⊂ b defines a as being a subclock of its superclock b. Every
instant of the subclock occurs synchronously with one of the instants of the
superclock: (∀k ∈ N? )(∃i ∈ N? )(a[k] ≡ b[i]).
Lazy Parallel Synchronous Composition of Infinite Transition Systems 135
Exclusion a # b. The clocks connected with this relation cannot have coinci-
dence instants: (∀k ∈ N? )(∀i ∈ N? )(a[k] # b[i]).
Precedence a ≺ b is the index-dependent relation. This operator is un-
bounded. Every instant of clock a has to precede the instant of clock b with
the same index (∀k ∈ N? )(a[k] ≺ b[k]).
Strict Precedence a ≺ b. This relation is a severer version of the previous
one in the sense that the instants of the clocks a, b with the same indices cannot
be equal: (∀k ∈ N? )(a[k] 4 b[k]).
4.2 The kernel expressions
A ccsl specification consists of clock declarations and conjunctions of clock
relations between clock expressions. A clock expression defines a set of new
clocks from existing ones. Most expressions deterministically define one single
clock. Table 2 gives a list of the ccsl kernel expressions.
Table 2. Basic expressions defined in the ccsl kernel.
Ref Name Kind of expression Notation Textual form
E1 Inf Mixed, unbounded a∧b a glb b
E2 Sup Mixed, unbounded a∨b a lub b
E3 Defer Mixed a ( ns ) b a def erred b f or ns
E4 Sampling Mixed a 7→ b a sampling b
E5 Strict sampling Mixed a→b a strictlySampled b
E6 Intersection Synchronous a∗b a clockInter b
E7 Union Synchronous a+b a clockU nion b
E8 Concatenation Synchronous a•b a f ollowedBy b
E9 Waiting Synchronous a fn b a wait n b
E10 Preemption Synchronous a b a upto b
(UpTo)
136 Yu. Romenska and F. Mallet
Sampling a 7→ b. The sampling expression ticks in coincidence with the tick
of the base clock immediately following a tick of the trigger clock and after it
dies. In the considered case, the trigger clock is b and the base clock is a. The
textual syntax of this expression is represented as c = a sampling b. In Figure 1
the automaton is given, where input symbol c is equal to the result clock of the
expression. The notation {a, b, c} denotes that the automaton remains in state
s2 if a, b and c tick all together simultaneously. if b and c tick simultaneously
without a then, the automaton goes back to state s1 . If a ticks alone, it stays in
s2 . All other cases are forbidden by the semantics of the operator.
{b} {a} {a}
s1 s2 {a, b, c}
{b, c}
{a, b}
Fig. 1. The automaton for sampling expression
Strict Sampling a → b. The expression is a strict version of the previous
one where c is emitted when the automaton is in state s1 , and a and b tick
simultaneously.
Waiting a fn b. The resulting clock ticks only once after a special number
given as a parameter of the base clock, and then the resulting clock dies. c = a
wait n b, where n is a given parameter (it is a natural number).
Preemption (UpTo) a b. The resulting clock ticks in coincidence with a, it
dies as soon as b starts to tick: c = a upto b.
Union This expression is non-terminating and index-independent. Its result is
a clock with set of instants which is a union of the instants sets of the clocks-
parameters that participate in the expression: c = a + b.
Intersection The result of this index-independent expression is the clock which
ticks each time when the clocks-parameters tick simultaneously: c = a ∗ b.
Concatenation a • b . The expression is terminating. The resulting clock ticks
in coincidence with the first clock-parameter a. After death of a it starts to tick
in coincidence with the second clock-parameter b . It should be noted that this
expression is valid only if the first clock eventually dies, i.e. a ↓ is specified.
Lazy Parallel Synchronous Composition of Infinite Transition Systems 137
Defer (Delay) a ( ns ) b. The parameter of the expression are a (the base
clock), b (the delay clock) and ns that represents a sequence of elements from
N>0 . The sequence of the natural numbers can have an infinite part. Let ns[i] be
the ith element of the sequence. We assume that if i > 0 then ns has an infinite
part, if 0 6 i < p there are p elements in the sequence. Every tick of the base
clock starts up the counter for respective element of the sequence. For every tick
of the delay clock the relative counter is decreased. When the counter reaches
1 the respective instant of clock b occurs. The textual form of the expression is
c = a def erred b f or ns.
Sup a ∨ b. The expression is index-dependent. The expression a ∨ b defines
a clock that is slower than both a and b and whose k th tick is coincident with
the later of the k th ticks of a and b. The formal definition is presented as:
(a 4 (a ∨ b))(b 4 (a ∨ b))(∀c ∈ C) : (a 4 c)&(b 4 c) ⇒ ((a ∨ b) 4 c). This
is a typical example of unbounded transition system (with an infinite number of
states).
{a, b} {a, b} {a, b, c} {a, b} {a, b}
{a, c} {a, c} {a} {a}
... s−1 s0 s1 ...
{b} {b} {b, c} {b, c}
Fig. 2. The automaton for sup expression
Inf a ∧ b. This is index-dependent and unbounded. It is the dual of the previous
one. The result of the expression is the slowest of faster clocks. It means that
the defined clock is faster than both a and b , the k th tick of this clock occurs
in coincidence with the earlier of the k th ticks of both a and b. ((a ∧ b) 4
a)((a ∧ b) 4 b)(∀c ∈ C) : (c 4 a)&(c 4 b) ⇒ (c 4 (a ∧ b)) is the formal
definition.
4.3 Unbounded CCSL operators
Lazy evaluation or call-by-needed is an evaluation strategy that delays the eval-
uation of an expression until its value is actually needed. This approach allows
construction of potentially infinite data structures and avoids repeated evalua-
tions. To construct the algorithm of the parallel execution of several automata, it
is necessary to have the possibility to work with infinite data structures (transi-
tion systems with an infinite number of states). Lazy evaluation provides the ap-
paratus for this task. The ccsl has four basic unbounded operators which can be
138 Yu. Romenska and F. Mallet
represented as infinite automata: the precedence relation, the strict precedence
relation, the inf expression (the fastest of slower clocks) and the sup expression
(the slowest of faster clocks).
Example 1. Let us consider as an example the automaton for the strict
precedence relation (Fig. 3).
Fig. 3. The automaton for strict precedence relation
If we only consider the possible outgoing transitions of each state, we can
distinguish the initial state (0 on Fig. 3) from the other ones. In the initial state,
only a can tick and must tick alone. In the other states, if a ticks alone, we must
progress to the right (increasing an unbounded counter), if b ticks alone, we must
go back to the left (decreasing the counter). If a and b tick simultaneously, we
remain in the same state. We assume that each state has a default transition to
itself, when nothing happens.
5 The synchronized product: an algorithm
We assume that the input for the algorithm is a given finite set of automata;
each of them can be either finite or infinite. The output is the automaton that is
the composition of the input ones. We denote the resulting automaton through
the tuple R. Let us introduce the set St which is the set of all unconsidered
states of the resulting automaton. At the beginning of the algorithm execution,
the initial states of input automata are considered as the current ones. The
first unconsidered state of the resulting automaton is the state formed from
current ones. For each considered current state of input automata the set of
available transitions from the given state is calculated. For all input symbols
of each of the calculated transitions the consistent solution is computed, which
is the set such that every original symbol is its subset. If the solution exists,
the new state is calculated from the states of input automata, in which we can
go on the appropriate input symbols that are the subset of the found solution.
If the computed new state does not belong to St, it is added to the set of
unconsidered states. The resulting automaton is completed by the new state
Lazy Parallel Synchronous Composition of Infinite Transition Systems 139
and the appropriate transition, which has the input symbol that is equal to the
respective solution. If the solution cannot be found, then we take a state from
St, in the input automata the appropriate states are considered as current. The
process is repeated while the set St is not empty.
5.1 The Formal Definitions
We introduce the formal definitions that are used to describe the algorithm.
C is the (finite) set of clocks and —C— is its cardinality;
state : C → StateDomain = {ST ALL, T ICK, DEAD} that denotes the state
of a clock.
A = {Ai }: the ordered set of input automata for the composition.
Ai = (Ci ; Si ; Ii ; moveOni ; si0 ∈ Si ; curi ∈ Si ; availT ransi ), where
Ci ⊂ C: the ordered set of clocks of the ith automaton;
Si : the set of states of the ith automaton;
si0 ∈ Si : the initial state of the automaton;
moveOni : Si × 2Ci → Si is the transition function;
curi ∈ Si : the current considered state;
Ci
availT ransi : Si → 22 : gives a set of available configurations for a given state.
The resulting composite automaton can be defined as follows:
R = (C; SR ; moveOnR ; s0R ∈ SR ; curR ∈ SR ; availT ransR ), such that:
C: the set of clocks of the composite automaton is equal to the set of global
clocks;
SR = {sR : sR = (s1 , ..., s|A| ), si ∈ Si , ∀i = 1...|A|}: each element of the set of
states of the resulting automaton consists of the states of input automata; the
element of the considered composite state corresponds to the automaton with
the same number;
moveOnR : SR ×2C → SR is the transition function for the resulting automaton;
s0R ∈ SR : the initial state;
curR ∈ SR : the current considered state;
C
availT ransR : SR → 22 : this function returns the set of input symbols for the
available transitions of the composite automaton.
The set of found solutions is represented by the set SOL = {sol : sol =
(st1 , . . . , st|C| ), sti ∈ {ST ALL, T ICK, DEAD}, ∀i = 1...|C|}.
St = {sR : sR ∈ SR } is the set of considered states of the resulting automa-
ton.
Get : Set → element, element ∈ Set: the function that returns an element
of the given set.
Register : C × {1..|A|} → 2C . This function returns an input symbol for the
th
i automaton based on the states of the global set clocks.
index : 2C × C → N, the position of a clock in an ordered set of clocks.
5.2 The Composition Algorithm
Global variables.
140 Yu. Romenska and F. Mallet
indexTable:=0: indexTable ∈ N;
array StateDomain temp[]: ∀i ∈ {0, ..., |C|-1}, temp[i] ∈ StateDomain;
array StateDomain cur_sol[]: ∀i ∈ {0, ..., |C|-1}, cur_sol[i] ∈ StateDomain;
array StateDomain OptionsTable[][]: ∀j ∈ {0, ..., |C|-1}, ∀i ∈ {0, ..., 3|C| -1},
OptionsTable[i][j] ∈ StateDomain;
array Boolean SolutionsTable[][]: ∀j ∈ {0, ..., |A|-1}, ∀i ∈ {0, ..., 3|C| -1},
SolutionsTable[i][j] ∈ {true, false};
cur_solution ∈ SOL
new stR ∈ SR
The main purpose of the following function is to build the composite au-
tomaton from the given set of the input ones. It uses three other functions:
buildSolutions(), getSolutions() and buildOptionsTable().
function composition(){
// the initial state is the cartesian product of all the initial states
|A|−1
1. s0R := s00 × ... × s0 ; St:={s0R };
3. while(St6= ∅){
4. curR :=GetElement(N);
5. SOL:=buildSolution(curR );
6. while(SOL6= ){ //for each solution
7. cur_solution:=Get(SOL);
// set clocks to the appropriate states
8. for(k:=0; k<|C|; k:=k+1){ C[k]:=cur_solution[k]; }
//form a composite state from the states of the input automata in which
// it is possible to go with the respective input sets
9. for(i:=0; i<|A|; i:=i+1){ new_sti := moveOn(curi , Register(C,i)); }
// include the new created state if it is not yet included
10. if(new_stR ∈/ SR ){
11. SR :=SR ∪new_stR ;
12. IR :=IR ∪cur_solution;
13. }
//include the created state in the set of unconsidered states
14. if(new_stR ∈
/ St){ St:=St ∪ new_stR ; }
//set the current states to the previous positions
15. for(i:=0; i<|A|; i:=i+1){ curi :=curR [i]; }
16. SOL:=SOL cur_solution;
17. }
18. N:=N curR ;
19. }
}
Lazy Parallel Synchronous Composition of Infinite Transition Systems 141
The following function calculates the set of possible solutions. It builds a table
of all options (OptionsTable) for the states of the clocks. The number of rows of
this table is equal to 3|C| . The base is equal to three because all three possible
states are considered (ST ALL, T ICK, DEAD). The number of columns is equal
to the number of the clocks in the global set. Also, the table for finding solutions
is defined (SolutionsTable). The number of rows is equal to 3|C| , the number of
columns corresponds to the number of input automata. If the input set defined
in a row of OptionsTable is available for the input automaton, then the value of
an element of SolutionsTable situated in the same row as the input set is set the
value true. When the SolutionsTable is completed (all available input symbols
of input automata have been considered), the function getSolutions() is invoked
to find the set of solutions SOL using the data of SolutionsTable.
function buildSolutions(sR ){
20. buildOptionTable(0);
//Initially there are no solutions
21. for(r:=0; r<3|C| ; r:=r+1){
22. for(i:=0; i<|A|; i:=i+1){ SolutionsTable[r][i]:=false; }
23. }
24. for(i:=0; i<|A|; i:=i+1){
// receiving the set of input sets for all available transitions
// of the current state of the appropriate ith
// automaton n is a number of available transitions
25. Iin := availT ransi (sR [i]);
//assignment of the value true to an element of SolutionTable
26. for(k:=0; k 0,
let us denote the ratio between Bn,k (m2 ) and Bn,k (m1 ) by a new function:
Bn,k (m2 )
tn,k (m1 , m2 ) = .
Bn,k (m1 )
Thus, by direct calculations we obtain the following formula for mutual in-
formation:
2 ln 2 − 1 2 ln 2 − 1 X 1 X
I(X(n) ; Θ) = − Bn,k (m1 ) − Bn,k (m1 )·
2 ln 2 2 ln 2 2
k∈An k∈An
tn,k (m1 , m2 )
tn,k (m1 , m2 )2 · log + log (tn,k (m1 , m2 ) + 1)
tn,k (m1 , m2 ) + 1
. (5)
1 − tn,k (m1 , m2 )
Let f (t) be a function defined as follows:
t
t2 · log + log(t + 1)
t+1
f (t) = .
1−t
Now it is easy to see that the formula (5) can be written as
2 ln 2 − 1 2 ln 2 − 1 X
I(X(n) ; Θ) = − Bn,k (m1 )+
2 ln 2 2 ln 2
k∈An
!
1 X
Bn,k (m1 ) · f (tn,k (m1 , m2 )) . (6)
2
k∈An
We claim that in the considered case (0 < m1 < m2 < 1) mutual information
is less than
2 ln 2 − 1
.
2 ln 2
To prove this fact it is enough to notice that the variable part of expression (6)
is always negative. Actually, we know that Bn,k (m1 ) > 0 and tn,k (m1 , m2 ) is a
Asymptotical Information Bound of Consecutive Qubit Binary Testing 171
ratio of two positive values, so, it is also positive. In addition, it is easy to show
in the classical way that for all t ∈ (0, 1) ∪ (1, ∞) function f (t) is greater than
zero.
Now we need to consider the special case, in which 0 < m1 < 1, m2 = 1. It
is evident that the case when m1 = 0, 0 < m2 < 1 is similar to the latter.
Suppose that 0 < m1 < 1, m2 = 1. Thus, on the one hand, for all k we have
Bn,k (m1 ) > 0. On the other hand, for all k < n Bn,k (m2 ) = 0, and only for
k = n Bn,k (m2 ) = 1. It is easy to see that in this case we obtain
2 ln 2 − 1
I(X(n) ; Θ) = +
2 ln 2
Bn,n (m2 )2 log Bn,n (m2 ) − (Bn,n (m2 )2 − 1) · log (1 + Bn,n (m2 ))
1
· =
2 Bn,n (m2 ) − 1
2 ln 2 − 1 1 2 ln 2 − 1
− · f (Bn,n (m2 )) < .
2 ln 2 2 2 ln 2
Thus, now we know that
2 ln 2 − 1
I(X(n) ; Θ) ≤ ,
2 ln 2
and, in particular, the equality is held if and only if the considered binary test
is projective.
4 Asymptotic Properties of Consecutive Measurements
In the previous section we have considered the case when the given qubit binary
test is a projective measurement. We have proved that only this type of measure-
ment allows to achieve the maximum of information about the initial state. As
far as the measurement is projective, repeating of the measuring procedure does
not provide any extra information. In addition, we have found the maximum
value of the accessible information:
2 ln 2 − 1
max {I(X(n) ; Θ)} = .
T∈M, n∈N 2 ln 2
In this section we return to considering of the general view of a qubit test,
and we work with consecutive qubit testing. So, this time we investigate the
dependence of the amount of information on n – the number of iterations. The
objective of this section is to prove that the maximum of accessible information
can be reached asymptotically by performing consecutive measurements using
an arbitrary qubit binary test.
More strictly, our aim is to prove the next theorem:
Theorem 1. Suppose we have a pure qubit state and we perform consecutive
qubit binary testing using the given test T = {M1 , M2 }. Then for arbitrary
172 A. Varava and G. Zholtkevych
ε > 0 there exists a corresponding number of iterations n(ε) such that for all
subsequent iterations (n > n(ε)) the following inequality is held:
max {I(X(m) ; Θ)} − I(X(n) ; Θ) < ε .
T∈M, m∈N
In other words, as far as the mutual information can be written as
2 ln 2 − 1 2 ln 2 − 1 X
I(X(n) ; Θ) = − Bn,k (m1 )+
2 ln 2 2 ln 2
k∈An
!
1 X
Bn,k (m1 ) · f (tn,k (m1 , m2 )) ,
2
k∈An
we need to find n1 (ε) such that for all n > n1 (ε)
X ε
Bn,k (m1 ) · f (tn,k (m1 , m2 )) < , (7)
2
k∈An
and n2 (ε) such that for all n > n2 (ε)
X ε
Bn,k (m1 ) < . (8)
2
k∈An
Therefore, for n > n(ε) = max{n1 (ε), n2 (ε)} both of these inequalities are held.
Let us fix a certain positive value of ε. At first we consider the left side of
inequality (7). Let us divide the set An into two non-intersecting subsets:
k
Γn (m1 ) = {k ∈ An } : − m1 < δ̃(m1 , m2 ) ,
n
k
∆n (m1 ) = {k ∈ An } : − m1 ≥ δ̃(m1 , m2 ) ,
n
where δ̃(m1 , m2 ) is a certain positive function.
It was demonstrated in [7], that for 0 < m1 < 1 and δ > 0:
X 1
Bn,k (m1 ) ≤ . (9)
4nδ 2
k∈∆n (m1 )
On the one hand, it is easy to see that f (t) is a bounded function. Suppose
that for all t ∈ (0, 1) ∪ (1, ∞) f (t) < C, where C is a certain positive constant.
Thus we have
X C
Bn,k (m1 ) · f (tn,k (m1 , m2 )) ≤ .
4n δ̃ 2 (m , m )
k∈∆ (m ) 1 2
n 1
So, we can choose a value n1,1 (ε) such that for all n > n1,1 (ε)
X ε
Bn,k (m1 ) · f (tn,k (m1 , m2 )) < .
4
k∈∆n (m1 )
Asymptotical Information Bound of Consecutive Qubit Binary Testing 173
On the other hand, we can see that when k is close to n · m1 , the value of
tn,k (m1 , m2 ) goes to zero as n goes to infinity. As far as lim f (t) = 0, there exists
t→0
a value n1,2 (ε) such that for all n > n1,2 (ε)
X ε
Bn,k (m1 ) · f (tn,k (m1 , m2 )) < .
4
k∈Γn (m1 )
Now let n1 (ε) be a maximum of values n1,1 (ε) and n1,2 (ε). Thus, for all n > n1 (ε)
inequality (7) is held.
Finally, let us consider inequality (8). Note that in the case of inequality
m1 < m2 the set An contains at most one element. Actually, by construction
k ∈ An if and only if Bn,k (m1 ) = Bn,k (m2 ). Solving this equation for the variable
k, we have
ln ((1 − m1 )/(1 − m2 ))
k0 = n · .
ln (((1 − m1 ) · m2 )/((1 − m2 ) · m1 ))
If n, m1 and m2 are such that k0 is an integer then An = {k0 }. If not, the set
An is empty. It is easy to show that lim Bn,k (m1 ) = 0, so, we can easily find
n→∞
n2 (ε) such that for all n > n2 (ε) inequality (8) is held.
Now let us build a rigorous proof of the considering statement using this
heuristic consideration. To do it, we firstly need several trivial propositions.
Proposition 1. The following statements are correct:
1. for all x ∈ (0, 1) the inequality x > ln (x + 1) is true;
2. for all x > 0 the inequality ln(x/(x + 1)) < −1/(x + 1) is true too.
This proposition can be easily proved using Tailor series expansion of ln (x) and
Euler’s transform applied to this expansion.
Proposition 2. Let x, y ∈ (0, 1) and x 6= y then the following inequality is held:
y (1−y)
x 1−x
<1.
y 1−y
The proof is omitted.
Now we can prove the above formulated theorem.
Proof (of Theorem 1). As we already know, the mutual information can be
presented as
2 ln 2 − 1 2 ln 2 − 1 X
I(X(n) ; Θ) = − Bn,k (m1 )+
2 ln 2 2 ln 2
k∈An
!
1 X
Bn,k (m1 ) · f (tn,k (m1 , m2 )) .
2
k∈An
174 A. Varava and G. Zholtkevych
We also know that
2 ln 2 − 1
max {I(X(n) ; Θ)} = .
T∈M, n∈N 2 ln 2
To prove the theorem it is enough to show that there exists n(ε) such that
for all n > n(ε)
X X
Bn,k (m1 ) · f (tn,k (m1 , m2 )) + Bn,k (m1 ) < ε .
k∈An k∈An
Consider an arbitrary ε > 0. Let us divide the set An into subsets Γn (m1 )
and ∆n (m1 ) in the following way:
k
Γn (m1 ) = {k ∈ An } : − m1 < δ̃(m1 , m2 ) ,
n
k
∆n (m1 ) = {k ∈ An } : − m1 ≥ δ̃(m1 , m2 ) ,
n
where δ̃(m1 , m2 ) is a certain positive function of m1 and m2 defined further.
Our aim is to prove that there exists such n(ε) that for all n > n(ε):
X ε
Bn,k (m1 ) · f (tn,k (m1 , m2 )) < , (10)
4
k∈∆n (m1 )
X ε
Bn,k (m1 ) · f (tn,k (m1 , m2 )) < , (11)
4
k∈Γn (m1 )
X ε
Bn,k (m1 ) < . (12)
2
k∈An
Firstly, let us consider inequality (10). We had already mentioned that for
all t ∈ (0, 1) ∪ (1, ∞) f (t) ≥ 0, and it is easy to see that for considered values of t
f (t) < 2 . So, using the above mentioned property of Bernstein basis polynomials
(9), we have:
X 1
Bn,k (m1 ) · f (tn,k (m1 , m2 )) ≤ .
k∈∆n (m1 )
2nδ̃ 2 (m1 , m2 )
Let n1,1 (ε) be sufficiently great. Then for all n > n1,1 (ε) the considering
inequality (10) is held.
Secondly, let us consider inequality (11). On the one hand, it is not hard to
find such δ(ε) that
t
t2 · log + log(t + 1)
t+1 ε
∀t ∈ (0, δ(ε)) : f (t) = < .
1−t 4
Asymptotical Information Bound of Consecutive Qubit Binary Testing 175
On the other hand, for sufficiently great values of n for all k ∈ Γn (m1 ) we have
tn,k (m1 , m2 ) < δ(ε). It follows that for great values of n we obtain
n
X ε X ε
Bn,k (m1 ) · f (tn,k (m1 , m2 )) < · Bn,k (m1 ) = .
4 4
k∈Γn (m1 ) k=0
√
1+(ε/2·ln(2))2 −1 1
Let δ(ε) = min ε/2·ln(2) ;2 . Then for t ∈ (0, δ(ε)) we have
ε · ln(2)
t< (1 − t2 ). As far as 0 < t ≤ 12 < 1, we obtain
4
2
t
t − t+1 t ε
= < .
(1 − t) · ln(2) (1 − t2 ) · ln(2) 4
If we combine this with Proposition 1, then for all t ∈ (0, δ(ε)) we have
t
t2 · ln + ln(t + 1) t + t 2
· − 1 t2
t+1 t+1 t − t+1 ε
f (t) = < = < .
(1 − t) · ln(2) (1 − t) · ln(2) (1 − t) · ln(2) 4
Now we need to find such n1,2 (ε) that for all k ∈ Γn (m1 ) : tn,k (m1 , m2 ) <
δ(ε). By definition,
k n−k
m2 1 − m2
tn,k (m1 , m2 ) = · .
m1 1 − m1
m2
As far as > 1, tn,k (m1 , m2 ) strictly increases with respect to k. So, if
m1
k
− m1 < δ̃(m1 , m2 ) then
n
m1 +δ̃(m1 ,m2 ) 1−m1 −δ̃(m1 ,m2 ) !n
m2 1 − m2
tn,k (m1 , m2 ) < · .
m1 1 − m1
Consider the right side of this inequality. Note that
m1 +δ̃(m1 ,m2 ) 1−m1 −δ̃(m1 ,m2 )
m2 1 − m2
·
m1 1 − m1
srtictly increases with respect to δ̃(m1 , m2 ). It is equal to 1 when δ̃(m1 , m2 ) =
δ̃ ∗ (m1 , m2 ), where
∗ ln ((1 − m2 )/(1 − m1 ))
δ̃ (m1 , m2 ) = − m1 .
ln ((1 − m2 )/(1 − m1 ) · m1 /m2 )
According to Proposition 2,
m1 1−m1
m2 1 − m2
· <1,
m1 1 − m1
176 A. Varava and G. Zholtkevych
we see that for δ̃(m1 , m2 ) ∈ (0, δ̃ ∗ (m1 , m2 )) we have
m1 +δ̃(m1 ,m2 ) 1−m1 −δ̃(m1 ,m2 )
m2 1 − m2
· <1.
m1 1 − m1
δ̃ ∗ (m1 , m2 )
Let δ̃(m1 , m2 ) = . Now it is easy to put n1,2 (ε) such that for all
2
n > n1,2 (ε) inequality (11) is held.
Finally, let us find such n2 (ε) that for n > n2 (ε) condition (12) is satisfied.
We have already seen that |An | ≤ 1, and the equality is held when
ln ((1 − m1 )/(1 − m2 ))
k0 = n · ;
ln (((1 − m1 ) · m2 )/((1 − m2 ) · m1 ))
is an integer. Let us denote the right side as n · c(m1 , m2 ) and write k0 as
k0 = n · c(m1 , m2 ).
Referring to the standard way of proving the Stirling’s formula, we can write
the following inequality:
√ n n √ n n
2πn · ≤ n! ≤ e · n · .
e e
It follows that
n e
r
n n k n n−k
≤ · · .
k 2π k(n − k) k n−k
So, it is now easy to see that
e
r
n n · m k n · (1 − m ) n−k
1 1
Bn,k (m1 ) ≤ · · .
2π k(n − k) k n−k
Substituting k0 = n · c(m1 , m2 ) for k in the last inequality, we get
s
e 1
Bn,k0 (m1 ) ≤ ·
2π n · c(m1 , m2 )(1 − c(m1 , m2 ))
c(m1 ,m2 ) 1−c(m1 ,m2 ) !n
m1 1 − m1
· . (13)
c(m1 , m2 ) 1 − c(m1 , m2 )
It follows from Proposition 2 that
c(m1 ,m2 ) 1−c(m1 ,m2 )
m1 1 − m1
· <1.
c(m1 , m2 ) 1 − c(m1 , m2 )
Now using (13) it is not hard to put n2 (ε) such great that for n > n2 (ε)
inequality (12) is held.
Finally, let n(ε) = max{n1,1 (ε), n1,2 (ε), n2 (ε)} . Now for all n > n(ε)
max {I(X(m) ; Θ)} − I(X(n) ; Θ) < ε .
T∈M, m∈N
The theorem is proved. t
u
Asymptotical Information Bound of Consecutive Qubit Binary Testing 177
5 Conclusions
In the paper the problem of obtaining classical information about the pure qubit
state using a single qubit binary test has been considered. It has been demon-
strated that the maximum of information is reached if and only if the using
measurement is projective. The maximum value of information has been calcu-
lated: n o 2 ln 2 − 1
max I(X(n) ; Θ) = .
T∈M, n∈N 2 ln 2
It follows, in particular, that to distinguish two arbitrary pure qubit states using
a single binary test it is necessary to have at least four pairs of qubits prepared
in the considered states.
It has been shown that the maximum of reachable information can be at-
tained asymptotically using an arbitrary consecutive qubit binary test. Thus, if
we have a certain measuring instrument performing a qubit binary test, we can
obtain an amount of information arbitrary close to the maximum.
As known [3, 6], Yu. Manin and R. Feynman proposed to use quantum sys-
tems to simulate others quantum systems. The results obtained in the paper show
that this idea should be refined: one should take into account all dependences
between an input data, a behaviour of a simulating system, and a structure of
an output data.
Our further research will deal with generalizing results of the paper for the
case of an n-level quantum system and a measurement with m outcomes.
References
1. Cover, T.M., Thomas, J.A.: Elements of Information Theory. John Wiley & Sons,
Inc. (1991)
2. Davies, E.B.: Information and Quantum Measurement. IEEE Trans. Inf. Theory
IT-24, 596–599 (1978)
3. Feynman, R.P.: Simulating Physics with Computer. Int. J. Theor. Phys., 21, 467–
488 (1982)
4. Holevo, A.S.: Bounds for the Quantity of Information Transmitted by a Quantum
Communication Channel (in Russian). Problemy Peredachi Informatsii, vol. 9, 3,
3–11 (1973)
5. Holevo, A.S.: Statistical Structure of Quantum Theory. Springer-Verlag, Berlin
(2001)
6. Manin, Yu.I.: Mathematics as metaphor: selected essays of Yuri I. Manin. AMS
(2007)
7. Natanson, I.P.: Constructive function theory. Vol. 1, Ungar, New York (1964)
8. Nielsen, M.A., Chuang, I.L.: Quantum Computation and Quantum Information,
10th Anniversary edition. Cambridge University Press, Cambridge (2010)
9. Shannon C.E., Weaver W.: The Mathematical Theory of Communication. Univer-
sity of Illinois Press, Urbana (1949)
10. Thawi M., Zholtkevych G.M.: About One Model of Consecutive Qubit Binary
Testing. Bull. Khark. Nat. Univ., vol. 890, 13, 71–81 (2010)
A Data Transfer Model of Computer-Aided Vehicle
Traffic Coordination System for the Rail Transport in
Ukraine
Denis B. Arkatov1
1
National Technical University "Kharkov Polytechnic Institute", Kharkov, Ukraine
denarkatov@gmail.com
Abstract. This paper gives a general layout of subsystem operation used for the
rolling stock traffic coordination. A principle of selection of information tech-
nologies applied for the realization of the communication and data transfer sys-
tem has been described. GSM and GPRS technologies that are used for the
transmission of navigation information on a vehicle locus and for the informa-
tion exchange in the system have been described. To provide a required level of
transmission capacity through the frequency reuse the guard period between the
base transceiver stations has been calculated. To calculate the GPRS channel
capacity and that of information exchange via the Internet GPRS network an al-
gorithm for the simulation modeling of packet arrival in the prescribed time
space has been constructed. Using this algorithm an experiment was carried out
for different intensity values of packet arrival to the system.
Keywords. Traffic coordination, information technologies, transmission capac-
ity, algorithm, simulation model, communication channel
Key terms. Development, Model, MathematicalModel
1 Introduction
The traffic control automation is a topical and up-to-date problem of the rail transport
in Ukraine. An important part of this problem is the development of algorithms for
the coordination and control of the large amount of rolling stock, situated in a zone of
railway traffic control points. An important requirement set to such algorithms is to
provide safe and regular traffic of the whole collection of trains and to make optimal
decisions from the economic standpoint.
To provide safe traffic for rolling stocks at the stage of departure or arrival a cer-
tain (permissible minimum) time interval should be provided, which is always taken
into consideration while making a train timetable. Despite this fact the train timetable
is sometimes disturbed due to many reasons and situations arise when the rolling
stocks are found to be undivided by the safe time interval and the groups of conflict-
A Data Transfer Model of Computer-Aided Vehicle Traffic Coordination … 179
ing trains are formed. In the case of high traffic intensity the groups of conflicting
trains at stations or at a train overtaking locus can be rather large. On the other part
the malfunction of train timetable can result in subsequent traffic disturbance (train
delay, train overtaking conflicts, etc.) The above problems can be resolved through
the automation of operative traffic control of rail transport. The experience gained by
foreign countries [1,2] shows that the efficient solution of traffic control with regard
to the rail transport is only possible if the latest information technologies are used.
The use of inexhaustible potential of railway information systems for the benefit of
entire transport system of the country allows to reduce control costs required for the
management and realization of domestic and international traffic. This also provides a
considerable improvement in the quality of transport and logistics-related services and
safety of traffic.
The examples of computer-aided systems that can fully or partially solve the above
problems are the European Train Control System (ETCS) [3], American Positive
Train Control (PTC) [4] and ILSD-U system [5] used by Russia.
A specific feature of the developed computer-aided system used for the rolling
stock (RS) traffic coordination is the introduction of contemporary satellite technolo-
gies, communication and data transfer systems into a routine work of rail transport in
Ukraine. The information exchange technology can be described as follows.
A GPS/GPRS –modem receives the navigation information from satellites, after
that using the GPRS technology the coordinates, current speed and other information
are transmitted via the operator’s server of mobile communication to the database
server. The RS locus data are transmitted to the workstation of railway station dis-
patcher and to the workstation of railway dispatcher.
The data should be processed in a real time mode. An important aspect is that a
time interval during which these data gain currency should be taken into considera-
tion. This fact considerably constrains not only the algorithm used for the solution of
coordination problem but also data transfer technologies.
This paper consists of several sections. First of all, the technologies used for the
data collection, processing and transfer will be described. Then, we will estimate the
channel data transfer capacity and do appropriate computations that prove the ob-
tained results.
2 Data Transfer, Processing, and Collection Technology
At the present time the railway adopts the computer-aided system (CAS), which in-
cludes the following basic subsystems: onboard intellectual system, which provides
positioning, control and information support for the rolling stock; surface intellectual
system (SIS), which provides control and coordination in a real time mode; communi-
cation and data transfer system based on the mobile GSM communication; navigation
charts that reflect a real railway infrastructure. The SIS structure and CAS system on
the whole are given in Fig.1.
A special place in SIS is occupied by the information system designed for the roll-
ing stock traffic coordination, which is the subject of this research. The navigation
180 D. B. Arkatov
data collection system is used for the automatic identification of a rolling stock locus
and also for securing the safety of traffic. The information about a rolling stock locus
is required for the optimal use of traffic and carrying capacity of railways and for the
elimination of dangerous situations (dangerous passing approach, passing the traffic
lights with forbidden signals, siding motion, exceeding the allowable speed in the
places of its restriction, etc.) A locus of each RS is transmitted to the traffic control
service via the navigation system for further data processing and traffic control. The
basic principles of monitoring are the safety of traffic, time fulfillment of a transporta-
tion schedule and costs minimization.
System of connection and data
Special navigation maps
transmission
Fig. 1. CAS and SIS structures
In order to create the appropriate computer-aided communication and data transfer
system for the rail transport the adoption of communication standard is required to
meet the entire system operation requirements, in addition to the use of satellite navi-
gation. The required communication system should provide high safety, reliability
and solutions proved in practice and it should also be an innovative and high perform-
ance system. The costs required for the total system deployment should be reduced to
a minimum.
A Data Transfer Model of Computer-Aided Vehicle Traffic Coordination … 181
Taking into consideration the European experience gained in the realization of
similar data transfer systems telecommunication technologies should be developed
meeting the GSM-R standard (Global System for Mobile Communications-
railway)[6,7], which engineering and economic efficiency has been proved not only
by tests but also by real application for different railway systems in developed Euro-
pean countries. However, at this stage of the development of information technologies
in Ukraine we can make to the conclusion that the realization of formulated problem
is possible and it is highly recommended to use the GSM mobile communication
standard. The GPRS technology is recommended for the data transfer.
The GPRS networks split the transmitted information into individual packets that
are delivered from a transmitter to a receiver. If errors have been detected the received
packets can be transmitted once again. The original message is designed by the re-
ceiver party using the obtained packets. The data transfer in packet- switching net-
works differs from the data transfer in channel-switching networks in the way that the
required channel resource is allocated exclusively for the time of transmission of ap-
propriate information packets. The rest of the time it is at disposal of a network. In
the case of GSM/GPRS networks this allows us to use one physical channel for the
transmission of packets to several subscribers and to simultaneously allocate several
physical channels for the transmission of packets to one subscriber. The packets are
transmitted aside from each other in different directions.
The GPRS defines the effective use of a channel resource. The same physical
channel is provided for the group of subscribers to receive messages sent from the
base transceiver station (BTS) to a mobile station (MS) and packets are transmitted as
soon as they arrive depending on the volume of information and the priority of sub-
scribers. Each packet contains an identifier or address, which is used for the delivery.
A subscriber is continuously connected to the packet network, which provides for him
a virtual channel. It becomes a real (physical) radio channel during the packet trans-
mission. The rest of the time this physical channel is used for the transmission of
packets of other users.
Due to the fact that the same channel resource is used by several subscribers and
during the communication session the packets of different users can simultaneously
arrive the waiting list of transmitted packets can be originated, which will result in the
communication delay. The allowable value of packets delay is one of the attributes
defining the quality of a subscriber service.
The paper [8] defines three classes of delay depending on the delay norms and
packet length. The top priority is assigned to the Class 1, the normal priority is given
to the class 2 and the class 3 enjoys the least priority. The delays have not been de-
fined for the Class 4, because the service of packets of this class is performed adher-
ing to the “best effort” principle.
The intensity values of packets arrival can be selected on the basis of statistical re-
search. A statistics shows that a number of packets arriving per time units to the input
of GPRS circuit changer can vary in a wide range from hundreds of packets /s at input
sites to several thousands of packets/s at backbone sites.
To provide the appropriate level of data throughput we use the basic principle of
the construction of cellular communication networks, which is based on the frequency
182 D. B. Arkatov
reuse [8]. The main essence of it consists in that the neighboring (adjacent) cells of a
mobile communication system use different frequency systems and in non-adjacent
cells located at sufficient distance from each other the used frequency bands are re-
peated. In practice the cities and regions with a solid cellular coating use clusters in
which each cell is divided into the three sectors, using the directional radiation an-
tenna with the directional pattern width of 120.
The base stations that allow frequency reuse are located at a distance D from each
other. This D distance is measured between the centers of hexagonal cells and it is
called a guard interval.
Proceeding from geometrical reasons the parameter D can be defined as follows:
D R 3 ,
where R is a radius of circle circumscribed around the regular hexagon; is a coeffi-
cient of the frequency reuse. The D ratio is defined as a reduction factor of channel
R
noises.
Thus, for the optimal frequency reuse and an increase in the GPRS channel capac-
ity in Ukraine the guard interval should be:
D 15 3 3 45 (km)
3 Case-Study for Capacity Evaluation
The developed computer-aided system used for the rolling stock traffic coordination
should take into consideration, while transmitting the navigation information to the
mobile operator server, not only the capacity of GPRS channels but also that of cable
or fiber-optic channel of the Internet network, which is used for the transmission of
data to the database server.
By analogy to the mobile communication it is necessary to consider a problem re-
lated to the average duration of delays in the packet switch.
The term “packet switch” implies here a concentrator (statistical multiplexor), a
virtual packet switch (network X*25, Frame Relay, and ATM network) and a router
(IP network).The packet switch can be represented as an element with many input and
output channels (switch/router). Using the Kendall notation such network elements
can be presented by queuing systems of G/G/1 or G/G/n type (arbitrary probabilistic
distributions describing the incoming stream of customers (in our case packets or
protocol blocks) and the time of their serving time. (Let us note that models with one
server, i.e. G/G/1 are often used for the analysis of packet switches).
Let’s assume that in the general case applications arrive to the system input in
compliance with the Poisson distribution law, whose service time is an arbitrary
value. Then the average queue length in the system with the infinite buffer size
(M/G/1) is calculated using the classic Khintchine-Pollaczek formula [9]:
A Data Transfer Model of Computer-Aided Vehicle Traffic Coordination … 183
q 1 C s2
tq t s 1 (1)
2(1 )
where q is an average queue length in the considered system (including protocol
blocks PB);
is the M/G/1 system load intensity, <1;
, are the intensity values of the PB arrival and service in the system, accordingly;
t s is an average time of PB service in the system;
Dt s
C s2 is a quadratic coefficient of service time variation equal to the ratio of
t 2 2
service time variance and squared expectation value.
The values required for computations using formula (1) were obtained through the
simulation modeling of data transfer in the Internet network, which algorithm is given
in [1].
A one-channel queuing system (QS) of the M/G/1 type has been taken as an exam-
ple. The arrival of the Poisson stream of applications with the constant service time
j has been simulated. The numerical experiment, which was carried out, allows us
to come to the conclusion that there is no loss of data packets.
The constructed graphs show the time of the arrival of customers in a queue and
the time of channel teardown from the previous application at the application arrival
1
intensities of 1 (see Fig.2.a) and (see Fig.2.b)
10 20
(a) (b)
Array 1 – The estimated time of the channel clearing
Array 2 – Actual time of the channel clearing
1 1
Fig. 2. QS simulation with an intensity of and
10 20
The given Figures show that an increase in intensity of arriving customers does not
result in the lack of channel capacity. This proves the correct choice of information
184 D. B. Arkatov
technologies for the realization of computer-aided system used for the rolling stock
traffic coordination.
To define the data transfer channel capacity in the Internet network the network
duration and delay should be calculated.
Due to the fact that the data transfer channel-switching center services packets it
can be simulated using the system with a constant service time of a M/D/1 type.
tq ts 1 .
21
At 1 and 1
10 5
1
2
Let t in the given example be equal to 0,05. As a result we get:
1
tq 0,05 1 0,075 ,
2
which corresponds to the first service class.
At 1 and 1
20 15
3
4
Let t in the given example be equal to 0,05. As a result we get:
3
tq 0,05 1 0,125 ,
2
which corresponds to the first service class.
4 Conclusions
Today the railways are the major branch of the economy of Ukraine and they serve as
a basis of the Ukrainian transportation system. Due to the rapidly changing demands
for freight services and carriage of passengers a permanent control over the required
amount of rolling stocks should be investigated. The proper amount of such vehicles
can be defined through the traffic analysis.
The information on the timetable of railway traffic and actual amount of rolling
stocks involved in traffic serves as a source information for further computations of
the capacity of a communication and data transfer system of the computer-aided sys-
tem. This will provide the fulfillment of actuality condition of obtained data and the
solution of the traffic coordination problem in a real time mode.
A Data Transfer Model of Computer-Aided Vehicle Traffic Coordination … 185
To provide the functioning of computer-aided system on the whole the following
algorithms should be realized. The algorithm used for the determination of the amount
of conflicting trains is the first step towards the problem solution. It allows for the
detection of those rolling stocks for which conflict-free conditions are not observed
due to different reasons (timetable violation, technical malfunction, etc.). A decompo-
sition algorithm is required for the reduction of dimension of a solved problem that
would allow the reduction of time interval of data acquisition. The obtained data are
used for the generation of control action and for the reduction of loading for the used
transmission channel, providing thus the observance of principles of control in a real
time mode. An algorithm of the problem related to the coordination of rolling stock
traffic allows for the elaboration of such control actions that provide not only the
elimination of a definite conflict but also the fulfillment of conflict-free conditions for
the definite rolling stock in the future. An algorithm of data transfer in a real time
mode is the basis of information exchange between the rolling stock and the database
server. Our further research is targeted at the development of algorithmic support of
communication and data transfer system to solve the problem on the coordination of
traffic of rolling stocks.
This paper gives the description of GSM and GPRS technologies that are used for
the transmission of navigation information on the locus of rolling stocks and also for
the information exchange in the system. It has been noted that in order to provide the
required capacity level we applied the main principle used for the construction of
cellular communication networks, i.e. the frequency reuse. The cluster structure with
a template of 3/9 has been constructed and the guard interval between the BSs used
for the mobile communication in Ukraine has been calculated.
To calculate the capacity of GPRS channels and that of information exchange
channel in the GPRS Internet network the algorithm of simulation modeling of arriv-
ing packets during the prescribed time interval has been constructed. On the basis of
this algorithm we carried out an experiment for different intensity values of packets
arrival to the system. Using the analysis data we can come to the conclusion that an
increase in the intensity of arriving customers causes no lack of channel capacity.
Our further research is targeted at the development of an algorithm for the data
processing in a real time mode, and also at the development and testing of the infor-
mation system designed for the rolling stock traffic coordination. We will also delve
into the computation of efficiency estimates to evaluate the introduction of informa-
tion technologies into the rail transport operation.
References
1. Arkatov, D. B.: Models and methods for automation of dispatching management for rail-
ways of Ukraine. Modeli i metodi avtomatizacii dispetcherskogo upravleniya dlya
zheleznodorozhnogo transporta Ukraini, Vostochno-Evropeyskiy zhurnal peredovikh
tekhnologiy, Eastern European journal of Enterprise Technologies, N 1/10 (61), pp. 61–63
(2013)
2. Arkatov, D. B.: Synthesis models of coordination of movement of mobile railway transport
of Ukraine. Sintez modeley koordinacii dvizheniya podvizhnikh sredstv
186 D. B. Arkatov
zheleznodorozhnogo transporta Ukraini, Vostochno-Evropeyskiy zhurnal peredovikh
tekhnologiy, Eastern European journal of Enterprise Technologies, N 4/3 (58), pp. 58–60
(2012)
3. ETCS requirements specification and validation: the methodology,
http://www.era.europa.eu/Document-Register/Documents/ETCS_methodology_v_1_2.pdf
(accessed 20 February 2013)
4. Positive train control (2013), http://en.wikipedia.org/wiki/Positive_train_control (accessed
20 February 2013)
5. Integrated locomotive safety device – unified, http://www.irz.ru/products/20/70.htm
(2012) (accessed 20 February 2013)
6. GSM-R (2013), http://en.wikipedia.org/wiki/GSM-R (accessed 20 February 2013)
7. GSM technical specification, http://www.ttfn.net/techno/smartcards/gsm11-11.pdf (1995)
(accessed 20 February 2013)
8. QoS in GPRS, http://doc.utwente.nl/18117/1/00000039.pdf (1999) (accessed 20 February
2013)
9. Tijms, H. C. A first course in stochastic models / Henk C., Tijms.p. cm. Includes biblio-
graphical references and index.
Quantitative Estimation of Competency as a Fuzzy Set
Leonid Vasylevych1 and Ivan Iurtyn1
1
Borys Grinchenko Kyiv University,
Department of Information Technology and Mathematical Sciences
lvasilevich@mail.ru, yurtyn@ukr.net
Abstract. The authors of this paper have used the assessment of competence as
a fuzzy discrete set consisting of essential capacities. There has been proposed a
procedure of competence quantitative estimation on the basis of discrimination
index of discrete fuzzy sets fixed on one totality. A linguistic variable “Compe-
tency coefficient” has been used here for making appropriate decisions on the
grounds of competency quantitative estimation. Assessment of a person’s com-
petency is proposed as a fuzzy discrete set consisting of necessary abilities as its
values. Using such competency assessment allows to estimate persons’ compe-
tency quantitatively and to compare them.
Keywords. Competency-oriented education, competence, capacities, fuzzy dis-
crete set, linguistic variable, membership functions, fuzzification, scalar capac-
ity of any fuzzy discrete set
Key terms. MathematicalModelling, MathematicalModel, FormalMethod
1 Introduction
The analysis of world education development tendencies demonstrates [1,3] compe-
tency-oriented education trend increase. Moreover, competency, which is not only
defined by knowledge, abilities, skills but also by considerably greater quantity of
factors (coefficients), becomes a major category both in education system and in the
job-market. Competency also includes the ability to obtain, to analyze and to revise
information; to learn through one’s lifetime; to change in compliance with the job-
market demands [1].
Thus, the quantitative estimation of competency necessary for making appropriate
decisions is a multicriterial problem, and therefore we need here to derive an integral
estimation of competency. Since there is no methodology of working out this prob-
lem, it makes the article topical for in it an integral index of a person’s competency
coefficient is estimated on the basis of the new competency assessment as a fuzzy
discrete set of which essential capacities are values.
Published works analysis. In the work [1], the key competencies concept has been
considered and three key competencies have been analyzed (specified by the Organi-
zation of Economic Cooperation and Development (OECD) representatives), which
are: autonomous activity; interactive facility use; ability to work in socially hetero-
188 L. Vasylevych and I. Iurtyn
genic groups. Federal Statistics Department of Switzerland and National Center of
Education Statistics of the USA and Canada within the program named Definition and
Selection of Competencies-- Theoretical and Conceptual principles (“DeSeCo”)”
summarized respective scientific results and different countries’ practices. In the work
[3] we give a review of works on the topic. But in all those works the qualitative ap-
proach to the named subject is solely used , but methods of quantitative competency
estimation have never been given.
Thus, the aim of this work is to develop methods of quantitative estimation of
competency on the grounds of its assessment as a fuzzy discrete set [2] consisting of
essential capacities.
Main results. Competency is defined in UNESCO publications as a combination of
knowledge, abilities, values and attitudes used in everyday life. Therefore qualitative
assessment of a person’s competency means his (her) ability to perform professional
duties or some functions efficiently. But this definition does not give a possibility to
estimate expert’s competency on quantitative basis. That is why we proposed to use
the following person’s competency definition.
Definition 1. A person’s competency is a finite discrete fuzzy set consisting of
abilities necessary for a job position or functions necessary for a respective position.
Membership functions of the set elements characterize the level of this competency
innateness to the person.
Definition 2. Abilities are necessary features, characteristics, faculties, qualities,
knowledge, techniques, skills and other traits which a person needs to perform duties
or functions at a respective position efficiently.
Thereby, in the beginning, we need to define at the discrete set of abilities
Y { y j : j 1, m} membership functions D yi 0 ; 1 of the fuzzy set D “Require-
ments necessary to perform duties or functions at a respective position efficiently”.
These membership functions characterize credibility, priority and importance of a
respective ability for a respective position or function.
Further we will use the notation of the discrete fuzzy set D in the form [2]:
Table 3. Designation D discrete fuzzy set
yi y1 y2 y3 … yn
D=
D(yi) D(y1) D(y2) D(y3) … D(yn)
or D y1 / D y1 ; y 2 / D y 2 ; y3 / D y3 ; ... y1 / D y n .
The set of abilities and respective membership functions will be different for each
position. When we specify the Y set we need to apply the Pareto principle, which
points that 20% of factors define 80% of the result. In practice, implementation of this
principle will lead to the effect that abilities with membership functions less than 0.5
will not be included into the D set. The task of specifying the set of abilities and
respective membership functions refers to the task of knowledge estimation by ex-
perts and demands creation of respective questionnaires.
Quantitative Estimation of Competency as a Fuzzy Set 189
As an example, let us specify an IT teacher’s information technology competency
in the form of a fuzzy set D:
Table 4. Example D representation of discrete fuzzy set
Yi y1 y2 y3 y4
D=
µD(yi) 1 0.9 0.7 0.8
in which y1 - ability to work in Word environment; y2 is the technique of work in
Excel environment; y3 is the technique of work in Excess environment; y4 is special
software skills (e.g, working out optimization tasks).
To perform a quantitative estimation of a particular teacher’s competency it is nec-
essary to estimate his abilities yi. To do so, tests, interviews, exams, respective lessons
control and other means can be recommended. Competency grades (their membership
functions estimation) can be shown on the scale from 0 to 1. This process is called
fuzzification. Hereby, for each person (teacher), we can define in the form of a fuzzy
set his personal fuzzy vector of abilities, which defines his competency. To define
μA(yi) a group of experts can be used who, after analyzing the person, answer the
question: “Is ability yi attributable to the person?” If the LD expert of L experts give a
positive answer, then
LD
μ А yi . (2)
L
As a rule this question does not have a single-value answer, so, experts can use
both binary logic (μAγ(yi) is either 0 or 1, where γ is an expert’s number) and fuzzy
logic (multiple-valued verity scale). In so doing they index the value of μAγ(yi)
0 ; 1 (subjective estimate). If quantity of the experts is L, then in the capacity of
μA(yi) we accept weighted arithmetic mean value of these estimates:
k А ( yi )
μ А yi L
, (3)
k
1
where kγ is the γ expert’s competency estimate.
For quantitative comparison of different persons’ competencies we need, firstly, to
compare in pairs finite discrete fuzzy sets D “Demands necessary to perform duties or
functions at a particular position efficiently” and Aj “the j person’s competency”
which are specified at one totality Y.
To compare these finite discrete fuzzy sets in pairs it is possible to use the estimate
P(D,Aj) of difference between D and Aj , which is reduced to the estimate of the trav-
erse of D Aj or D AJ [2]:
D Aj D (4)
РD , AJ ,
D
190 L. Vasylevych and I. Iurtyn
where the ... sign means scalar capacity of any fuzzy discrete set B [2]:
B μ B yi (5)
х Х
operation B of complementing the fuzzy set B is defined by the membership func-
tion [2]
μ _ y 1 μ B y , y Y (6)
B
operation of two fuzzy sets unification (C B K) has the membership function [2]
μ С y maxμ B y ; μ K y , y Y . (7)
In so doing P(D,Aj) as a rule is not equal to P(Aj,D). This attribute is used to com-
pare fuzzy sets specified at one totality: if P(D,Aj) > P(Aj,D), then the fuzzy set D <
Aj and vice-versa.
A person’s abilities, which have some membership functions’ value greater than
the value of respective abilities’ membership functions in the D set, must not compen-
sate small values of the Aj set membership functions. To avoid this it is necessary to
perform the Aj set normalization: membership functions values of the Aj set which
exceed respective values in the D set have to be equated to respective membership
functions’ values of the D set. Thereby it is necessary to insert the normalized fuzzy
set Ajн into the (3) formula.
Let us perform a comparison of two persons’ competencies. Let us define one per-
son’s competency by means of a fuzzy set А1 (y1/0.6); (y2/0.9); (y3/0.7); (y4/0.9)
and the other person’s competency by means of a fuzzy set А2 (y1/0.8); (y2/1);
(y3/0.5); (y4/0.9).
After normalization we have: А1н =А1 (y1/0.6); (y2/0.9); (y3/0.7); (y4/0.9);
А2н (y1/0.8); (y2/0.9); (y3/0.5); (y4/0.8).
The estimate of the difference P(D,A1) is equal to (3):
0,6 0.9 0.7 0.9 0.6
PD, A1 0.735.
1 0.9 0.7 0.9
The estimate of the difference P(A1,D) is equal to:
1 0 .9 0 .7 0 .8 0 .9
P А1 , D 0.806 .
0 .6 0 .9 0 .7 0 .9
We propose to calculate competency coefficient K is as the normalized estimate of
differences:
minP A, D ; PD, A (8)
K
P A, D
Quantitative Estimation of Competency as a Fuzzy Set 191
This coefficient always belongs to [0;1] interval. If P(A,D) > P(D,A) then K<1,
and if P(A,D) < P(D,A) then K=1.
After inserting computed estimates into the (7) formula we have:
min0.806;0.735
K 0.912 .
0.806
Calculation of competence coefficient K for the second person will give values
described below:
3.2 0.6 3.4 0.8
PD, A2 0.765 ; P A2 , D 0.813;
3.4 3.2
minP A, D ; PD, A min0.813;0.765
K 0.941.
P A, D 0.813
Thereby, we can conclude that the second person’s competency is greater than the
first one’s.
To define a person’s competency level basing on the competency coefficient value
it is necessary to specify a linguistic variable (LV) [2] “A person’s competency coef-
ficient”, which we will determine by means of a tuple
E , E j , j 1,5; E j ( x) 0;1; x K 0;1; .
Terms of “Competency” LV can be: E1 – very low competency; E2 – low compe-
tency; E3 – medium competency; E4 – high competency; E5 – very high competency.
Trapezoidal membership functions of terms can be defined by experts by means of
four numbers a : b : c : d , which define each term.
Using trapezoidal membership functions of terms and considering Harrington’s
scale it is possible to specify “Competency” LV as follows: E1 0 : 0 : 0.1 : 0.2 ;
E 2 0 .1 : 0 .2 : 0 .3 : 0 .4 ; E 3 0 . 3 : 0 . 4 : 0, 6 : 0, 7 ; E4 0.6 : 0.7 : 0,8 : 0,9 ;
E5 0.8 : 0.9 : 1 : 1 .
Let estimate Ej of a term by an γ expert amounts E j aj ; bj ; cj ; dj , then in
the capacity of membership function Ej of the term we accept a fuzzy quantity
1 L 1 L 1 L 1 L
Ej aj ; bj ; cj ; dj . (9)
L 1 L 1 L 1 L 1
To specify terms more appropriately Delphi technique can be applied.
Specifying membership functions lateral branches by straight line segments does
not reduce persons’ competency estimate’s generality but simplifies mathematical
operations over fuzzy quantities considerably [4]. In so doing the left l (x) and the
right r (x) lateral branches of the membership linear function have analytical form
respectively:
xa
l ( x) ; x a; b , (10)
ba
192 L. Vasylevych and I. Iurtyn
dx
r ( x) ; x c; d . (11)
d c
For the just made example, we have ascertained that the competence coefficient
K1=x=0.912 belongs to E5 term (very high competency) with membership function
(verity) one, and the competence coefficient K2=x=0.88 belongs to E4 term (high
competence with membership function 0.2 and to E5 term with membership function
0.8.
Algorithm of a person’s competency estimation consists of six stages: the prepara-
tory (1 to 4) and operational (5, 6) ones.
Y { y j : j 1, m}
1. Specifying the set of abilities for a position or functions.
2. Abilities’ membership functions assessment (Specifying D fuzzy set) ( (1) and (2)
formulae are applied).
3. Ai fuzzy set assessment – “A person’s competency”.
j E, E , j 1,5; j ( x) 0;1; x 0;1;
E
4. “A person’s competency” LV assessment: .
5. A person’s competency coefficient computing ((3); (4); (5); (6) and (7) formulae
are applied).
Computing a person’s competency coefficient’s membership functions to respec-
tive terms LV “A person’s competency” (formulae (9) and (10)).
At the preparatory stage experts are used, who define the notion “Competency” as
a discrete fuzzy set, the values of which are abilities necessary for a particular posi-
tion or functions.
Point 3 demands creating respective techniques, tests, problems and tasks that al-
low estimating various abilities of a person (to find membership functions of each
ability).
At the stage of receiving a person’s competency quantitative estimate points 5 and
6 are performed.
To specify A fuzzy set of an expert’s antecedent characteristics the expert’s ques-
tionnaire data, his (her) tests, interviews can be used.
The examined competency estimation methodology based upon using fuzzy sets
and a linguistic variable allows resolving several problems: conversion from current
qualitative competency assessments to quantitative estimation; multicriteriality of
competency estimation problem; impossibility of quantitative measuring certain par-
ticular indexes of competency; impossibility of real experiments to estimate different
persons’ competency.
2 Conclusions
1. Assessment of a person’s competency is proposed as a fuzzy discrete set consisting
of necessary abilities as its values. Using such competency assessment allows to
estimate persons’ competency quantitatively and to compare them.
2. A methodology of a person’s competency quantitative estimation is proposed.
Quantitative Estimation of Competency as a Fuzzy Set 193
3. It is proposed to estimate quantitatively persons’ competency on the basis of dif-
ference coefficient of finite discrete fuzzy sets D “Demands necessary to perform
duties or functions at a particular position efficiently”.
4. It is proposed to specify an expert’s competency coefficient in the form of linguis-
tic variable “Competency”.
References
1. Key Competencies: A Developing Concept in General Compulsory Education.
EURYDICE. The Information Network on Education in Europe, p. 224 (2002)
2. Pospelov B.A. (ed.): Fuzzy Sets in Management and Artificial Intelligence Models. Sci-
ence, Moscow (1986)
3. Sysoyeva S.O.: Education and Personality in Post-Industrial World. Monograph. KSPA,
Khmelnytsky (2008)
4. Vasylevych, L.F. Malovik, K.N., Smirnov, S.B.: Quantitative Methods of Making Deci-
sions in Terms of Risk. SNUNEP, Sevastopol (2007)
1.4 Methodological and Didactical Aspects
of Teaching ICT and Using ICT
in Education
New Approaches of Teaching ICT to Meet Educational
Needs of Net Students Generation
Nataliya Kushnir1, Anna Manzhula1 and Nataliya Valko1
1
Kherson State University, 27, 40 rokiv Zhovtnya St., 73000 Kherson, Ukraine
kushnir@ksu.ks.ua ilovetrees@mail.ru valko@ksu.ks.ua
Abstract. The paper describes the educational needs of modern students as
generation Net representatives, highlights the contradiction between their char-
acteristics and traditional ways of teaching ICT disciplines. The paper reports
teaching experience and poll results for three years at Kherson State University,
Ukraine. The purpose of the paper is to offer new teaching approaches to cope
with a generation gap and way to improve the quality of ICT teaching.
Keywords. Generation gap, Generation Net, teaching approaches, ICT disci-
pline
Key terms. ICTTool, TeachingPattern, TeachingMethodology, Capability
1 Introduction
The UNESCO report on Information Technology in Education [1] informs that
Ukraine is on the way of "the rapid advancement (progress) of ICT in education that
leads to the constant updating of the educational content and the quality of ICT train-
ing". However, there is a great amount of problems. Primary it is connected with the
fact that educational institutions and teachers in particular are not ready to the transi-
tion to the information society: “increased demands for flexibility, mobility and
adaptability to the education management system, educational institutions and teach-
ers in the context of rapid changes make it difficult to maintain and improve the qual-
ity of educational services."
2 Related Work
Using our teaching experience we identified the common trend of learning styles
among the students. Further research is required to investigate digital competence
formation among future teachers. Our attempt to find out an unknown factor that has a
significant impact on the pedagogical process and to understand the nature of this
phenomenon was based on the considering modern students as the representatives of
the new generation.
196 N. Kushnir, A. Manzhula and N. Valko
To date, about 40 books and scores of articles and papers have been written on this
generation that report the results of international surveys and other research and de-
scribe their characteristics. Their impact on education at all levels has been of major
interest to researchers and educators. There are about 10 terms to describe the current
generation of students [2]: Millennials (Howe and Strauss), Generation Y or Gen Y
(Nader), Echo Boomers (Tapscott), Net Generation (Tapscott), Digital Aboriginals
(Tarlow and Tarlow), Digital Natives (Prensky), Nexters (Raines and Filipczak), Dot
Com Generation (Stein and Craig).
The representatives of this generation were born in 1982-2003. Today’s students
and post-graduates are aged from 10 to 30. It means that all teachers and institutions
that are involved in the education of the students who have grown up in the world of
new digital, mobile and high-tech, digital technologies.
The technology itself has had a profound effect on this new generation, unlike on
any previous one. In the classroom, students can chat on Skype or write SMS to their
friends, take notes on ipad, surf the Internet and read a book on the ReadBook. This
behavior can not be fully appreciated by their teachers: it’s considered that electronic
instruments and digital devices distract students from the "real" study [2]. Majority of
today’s teachers are representatives of the previous generations. They are using learn-
ing models fitted for the teachers themselves but not for the new generation of stu-
dents.
It was found that representatives of the new generation inherent a wide range of
characteristics that are defined a predisposition for becoming a successful educator.
Realizing their significance for the education and considering themselves to be an
instrument of world changes, they will be strongly motivated to improve the quality
of life of the society. Some research study and develop strategies for a retention them
in definite professional sphere including education.
Moreover, the representatives of Net generation have solid moral values connected
with a family and our society. They are highly motivated to create an open and toler-
ant society. It’s important for new gens’ educators to consolidate and maintain
youths’ system of values. [15].
The results of the 3rd year students’ polling showed the low level of professional
awareness (Preschool and Elementary School faculty). Only 50% of students have an
intention to become teachers [16]. The lack of professional focus among students sets
a hard task for educators to make teaching a valuable and desired profession.
Among the characteristics of generation should be noted that Gen Y workers are
usually educationally focused and attribute their success to their educational capabili-
ties. They want to have successful careers. They do not like the dress code, demand
ICT equipped workplaces and want have a flexible work schedule.
Ronald A. Berk synthesized pertinent research evidence based on ten national and
international surveys: EDUCAUSE [4], College Students‘ Perceptions of Libraries
and Information Resources Survey, Greenberg Millennials Study [5], Education Re-
search Institute (UCLA) [3] American Freshman Survey [11], National Center for
Education Statistics [9], Net Generation Survey [8],The Net Generation: A Strategic
Investigation [13], Nielsen Net View Audience Measurement Survey [2, 10], Pew
Internet and American Life Project [6, 7] и Technological preparedness among enter-
New Approaches of Teaching ICT to Meet Educational Needs … 197
ing freshman [12]. The research results from the surveys and aforementioned books
has yielded twenty learner characteristics typical for most Net Geners: technology
savvy, relies on search engines for information, interest in multimedia, create Internet
content, operate at ―twitch speed, learn by inductive discovery, learns by trial and
error, multitask on everything, short attention span, communicate visually, crave so-
cial face-to face interaction, emotionally open: Embrace diversity and multicultural-
ism, prefers teamwork and collaboration, strive for lifestyle fit, feels pressure to suc-
ceed, constantly seeks feedback, thrives on instant gratification, respond quickly and
expect rapid responses in return, prefers typing to handwriting. The research results
from the surveys and aforementioned books have yielded twenty learner characteris-
tics typical for the most Net Geners.
Fig. 1. Improvement discipline background
We have identified some teaching approaches that are contradictory to the contem-
porary students’ needs. This gap is especially obvious in teaching computer related
disciplines. Complete comprehensive step-by-step instructions and exclusively indi-
vidual learning are no longer efficient. This context led educators to a revision of
present teaching strategies.
3 Setting up the Pedagogical Experiment
The only discipline - “New Information Technologies and Technical Facilities of
Education” – that concerns the formation of ICT skills is taught for the future teachers
of all specialties at Kherson State University. This year the discipline was renamed in
"Information Technology" in most curricula.
The teaching experience of teaching the discipline "New Information Technologies
and Technical Facilities of Education” in 2011 (109 students) and in 2012 (112 stu-
dents) at Faculty of Pre-School and Elementary Education (FPEE) allowed us to iden-
tify problems that are mainly related to the mentioned contradictions [17].
198 N. Kushnir, A. Manzhula and N. Valko
Table 1. The contradictions of the present ICT teaching approaches to the generation Net char-
acteristics
ICT teaching approach Generation Poll results
Net charac-
teristic
ICT teaching "from scratch": dis- Tech savvy 84% of respondents have started to use
regard (neglect) the actual level the computer for learning 7 years ago or
of the student's ICT skills. As a earlier
result, school ICT discipline as-
signments are duplicated, lab
manuals are detailed and tend to
be comprehensive.
Teaching materials are not inter- Relying on About 26% of the students classified
active and update search (attributed, mentioned, placed, noted)
engines for search engines to (as) the most fre-
information quently used sites on the Internet
Weak level of visualization of Interest in Movies and computer games have the
teaching materials, lack of inter- multimedia, second position among the purposes of
activity including hypertext "visual" using computer ranked by students
communica-
tion
Step-by-step manuals that pre- Creation of Social nets, wiki-sites and forums have
suppose learning by copying the Internet the third position among the most fre-
sample, the absence of the origi- content quently visited sites on the Internet. All
nal product as a result of the stu- of them are a platform for a creation of
dents’ work. their own content, express their opinion,
share things made by themselves. 13% of
respondents mentioned the creative ac-
tivity as a major purpose for using the
computer
Students are constrained by one Multitasks The sum of hours for different every-
plotline (storyline) in learning, on every- day life activities informed by student
the absence of immersion, prob- thing are about 28 hours a day.
lem-solving and decision-making
tasks and enough freedom for
actions in realization students’
learning trajectory.
Weakly realized person-centered Emotionaly Social nets were ranked as the second
approach open position by students
Individual fulfillment and per- Teamwork Using computer for communication
formance of learning results and coop- among students is placed on the third po-
eration sition after learning and entertainment
purposes by students
We interviewed students of Faculty of Pre-School and Elementary Education in
2012-2013 academic year. The results confirm last year statistics that quality of
teaching materials on KsuOnline was highly appreciated by students, average score
was 9.29 in 2011 and 9.16 in 2012 out of 10.
New Approaches of Teaching ICT to Meet Educational Needs … 199
The results of the entrance poll of other faculties in 2011-2013 academic years
showed the following trends:
In 2011, 89% of the respondents owned a computer or a laptop. This academic
year, 100% of students are the owners of a computer or a similar device regardless
of the discipline.
70% (2011), 68% (2012) and 69% (2013) students have an access to the Internet
outside of the university – thereby, this rate stays the same.
However, a number of students who recognize themselves addicted to the Internet
has grown from 24% in 2011 to 32% in 2013.
An interesting result was found by visiting University website by students from
different disciplines. 17% of the third year students said they had never visited the
university website at the departments where teachers hardly use ICT in teaching.
On the other hand the rate was 0 (since 2011 to date) at the departments where
most of the teachers regularly use ICT.
The number of students who have a positive attitude towards the use of ICT in
education (inter alia, at lectures) has increased (see Fig. 2).
Fig. 2. Students Attitude Towards the Use ICT During Lectures
The purposes of using the computer by students have almost the same rating,
regardless of year and the faculty (see Fig. 3).
200 N. Kushnir, A. Manzhula and N. Valko
35%
30%
25%
20%
15%
10%
5%
0%
To do my lessons To communicate by the To learn something new To entertain themselve I have no computer
Internet (distance learning (games, watching
courses, video tutorials, movies, etc.).
workshops, etc.)
FFP, FT, FPHS (2012‐2013) FPSEE (2012‐2013) FPSEE (2011‐2012)
Fig. 3. The Purpose for Using a Computer and Internet by Students
Social net is becoming more popular among services for communication (see Figure),
IME and email are also frequently used.
Fig. 4. Student Communication Services
Modern students overestimate their skills for online searching according to Berk.
The results of our research have confirmed this fact. The results of entrance poll and
test in Informatics are contradictory. The majority of students estimated their level of
ICT proficiency as excellent and good. The rate is depicted below.
New Approaches of Teaching ICT to Meet Educational Needs … 201
Fig. 5. Results of self-appraisal a computer user by students (FFP, FT FPHS, 2013)
The entrance test contains 45 comprehension questions and is designed by analogy
with ECDL test. The aim of testing is to check residual knowledge of school
computer discipline. The results showed that only 3.28% of students had a score
higher than 4.5, 82% had a score higher than 3 but less than 4,42 points, 62% had
from 2 to 3 points and 1.64% scored less than 2 points.
During the recitations, the following problems were identified that are related to
organizational issues and students attitudes:
Students have no skills to manage their time and work efficiently without strict
teacher control, regardless deadlines. Only few students uploaded their works in
time.
Not all students use email regularly. Many of them chronically forget their email
user names and passwords. Students are confused by various browsers or other
version applications. As a result, the login process to email or distance learning
system regularly takes up to 7 minutes.
There is no clear understanding of terms among students. Being able to work with
some programs, often they cannot deal with similar ones.
It is necessary for a teacher to consider different speed of students work while
planning lessons. Some students do not regularly use computers. Thus, typing,
mouse control, and searching for a command causes a delay.
Plagiarism. Some students prefer not “to waste” their time to do lessons. They
easily copy their colleagues’ results or download similar ones from the net. It is
important at the first lesson to highlight the value of students own creative work
and punish the students who attempt to copy someone else work. Under strict
control, the number of plagiarisms decreases, but does not disappear. In fact, we
notice that plagiarism is growing with task level. The reason is that the students
desire to get a mark, not knowledge.
The students do not like reading long instructions. Usually they prefer to ask a
teacher or colleagues than to read step-by-step manual. It is also associated with
different systems prevailing perception among students. The feature of the course
202 N. Kushnir, A. Manzhula and N. Valko
is that the most of the teaching materials are stored in digital form (in digital form).
Creating e-learning course does not require additional financial costs, as opposed
to hard copy, faster and easier to edit and update. However, reading from the
screen does not give effective results. Perceived only about 40% of the
information.
Lack of collaborative activities: in pairs, groups and teams, social assessment,
problem-solving and decision-making tasks. A student presents results of an
individual work only to one person–his teacher, and feels subordinated under such
conditions. In addition, it is the reason of the absent of the critical view on his own
work. This results in frustration and dissatisfaction of the student to any criticism.
The teacher comments are perceived in a negative or indifferent way.
Educators should pay special attention to students feedback about effectiveness of
teaching activities and the level of their satisfaction. For example, monitoring and
evaluation allow a teacher to obtain an information about teaching incomes and
quality, that would improve his work. It should reflect the level the students
knowledge. It’s vice versa in practice: evaluations become a sign of knowledge /
ignorance of the material, the student rating of the group.
The presented requirements to discipline designed in previous work on digital lit-
eracy formation [17] have been adapted to Berk’s pedagogical strategies. As a result,
we have formulated new approaches considering educational needs of the Net stu-
dents’ generation (see Table 2).
Table 2. The approaches of teaching ICT discipline to meet educational needs of generation
Net students
The teaching element Requirements (Description)
Discipline content A discipline content should reflect current research and encour-
ages students to use new approaches and technologies, highlights
current trends in ICT development.
Tasks presuppose creative activity, form skills of self-learning
and further development. Examples are inspiriting.
Task execution result is useful, valuable and applicable product in
professional practice.
All elements of the discipline are focused on future professional
activities.
Students motivation All elements of the course (assignments, surveys, etc.) help
students to see themselves in their future profession, particularly
in teaching profession.
It is important to emphasize interest to the students’ opinion, to
make possible to their contribution in doing collective projects,
for example, to create a bank of teacher’s materials;
Student wants to get a feedback from his colleagues about his
work. Any result of creative task should pass following stages:
creation, publication and social assessment with definite criteria.
Student wants to evaluate teacher’s work too, so a teacher should
organize the ways to impact on the discipline development for
students, to express their wishes.
New Approaches of Teaching ICT to Meet Educational Needs … 203
The teaching element Requirements (Description)
Organizational issues Tasks presuppose collaboration, facilitate communication and
interaction to develop personal aspect of student and help to
realize his individuality.
Clear structure, planning, to do list and deadline system.
Active use of formative assessment techniques.
Ice-breaking, team-building and communicative exercises at the
beginning and at the end of the lesson to form communicative
skills, values, relationship.
Use no more than two new environments at the lesson.
Teacher should consider in the selection of services to work at the
lesson:
registration absence or its simplicity,
functionality
easiness,
necessity to install additional software
opportunities to use in profession.
The main aim of updating the discipline was to help future teachers to create and
organize their own learning space in the Internet. Therefore online interactive services
that can be used for communication and teaching pupils were included (creating word
clouds, mind map, open online documents, site, etc.). We also considered Berk’s
strategies while designing the discipline. The updated version of "Information Tech-
nology" discipline was taught at the Faculty of Foreign Philology (FFP) - 104 stu-
dents, Faculty of Translation (FT) – 61 students, Faculty of Psychology, History and
Sociology (FPHS) – 28 students.
Table 3. The implementation of Berk’s teaching strategies in the course "Information Technol-
ogy"
Characteristic The implementation of educational strategy
of Net Generation
Tech savvy The virtual discipline environment was created with the system of
distance learning ksuonline.ksu.ks.ua located on a MOODLE plat-
form. This system allows developing a course with such elements as
glossaries, wikis, multimedia clips, presentations, tests, blogs, fo-
rums, etc.
Teamwork and coop- Some tasks are complex and presuppose collaboration, while their
eration execution the cognitive interpersonal communication and interaction
of all participants progress. An important stage is to prepare a group
of students to work together. To ensure about students’ readiness for
cooperation teacher should arrange Ice braking, team-building and
communicative exercise (up to 5 minutes) at the beginning and at the
end of lesson. Teacher, who works in the computer classroom, usu-
ally recognizes his students primarily "from the back." Therefore,
such exercise facilitates personal contact with the group, which is
especially important while teaching short disciplines.
Interested in multime- The discipline content was developed and designed considering
204 N. Kushnir, A. Manzhula and N. Valko
Characteristic The implementation of educational strategy
of Net Generation
dia, "visual" communi- students' interest in "visual communication" and multimedia. Video-
cation clips and presentations were added to each theme. The tasks and
manuals contain minimum of text information and maximum of
graphics and illustrations. Teachers included such elements as blogs
and forums.
Rely on search engines. In adding to use search engines during the course students active
Multitasks on every- work with Google services as Google.Drive, Google.sites, etc. This
thing satisfies the interest in creating online content and allows simultane-
ously work with several documents. It’s also an efficient tool of
collective activity organization.
Retention in the profes- Organization of students’ activity is a process of consistent model-
sion ing of professional activity of specialists under learning conditions.
The students of teaching specialties create educational games, tests,
documents, tables that are useful in their professional practice. The
results of the work are evaluated not only technical element, but also
educational one. Also such task as creating a presentation "My
choice" (students aimed to describe situation of their professional
choice, analyze their present achievements and capacities, goals and
dreams, ways to attain them) and collective writing of mind map
"Modern teacher” were added.
Emotionally open Tasks are person-centered and presuppose creation unique results
that will describe student’s own lifestyle, attitude and etc. This ap-
proach helps decrease the quantity of plagiarisms and contributes to
the development of their creative abilities.
Each student’s work passes through formative assessment: self-
appraisal, social assessment, teacher’s assessment.
Such tasks as making clouds of words, mind maps, playing learning games are
relevant to professional and educational discipline orientation. In addition, they
presuppose creative activity and can be easily adapted to a group work.
Poll Results of 2012-2013 academic year at Faculty of Pre-School and Elementary
Education are comparable with poll results published one year earlier. The quality of
teaching materials on KsuOnline highly appreciated by students, average score was
9.29 in 2011 and 9.16 in 2012-2013 out of 10. The novelty of lectures was assessed as
7.59 in 2011 and 6.84 in 2012-2013, the novelty of practical tasks had 7.64 and 7.39
respectively.
The students’ preferences about the most interesting and useful for future
professional tasks have changed. In both of these categories the task "Creating
didactic game" leads. Making a poster "It's interesting to know," creation of "Site
Class" and creation of online poll follow in the rating. These tasks we preserved in the
updated version of the discipline "Information Technology". As a result, new version
of "Information Technology" discipline has the following structure:
Lesson 1. Registration on KSUOnline, taking an entrance poll and test.
Presentation "My choice" with the following structure:
1. My strengths and weaknesses (academic, creative, personal, individual aspects).
New Approaches of Teaching ICT to Meet Educational Needs … 205
2. My goals, objectives and deadlines (to 2020 year).
3. My completed tasks.
4. A letter from You yourself in 2020.
Lesson 2. Creating a class site on Google Sites (Students add, delete pages, edit
them and insert pictures and text, change site’s design). Taking a poll “NET Gener
profile Scale".
Lesson 3. Creating a poll on Google Drive and inserting it to the site.
Lesson 4. Creating a cloud of words (Tagxedo service), publishing pictures with a
cloud of words and writing an assignment for the pupils on the "Homework” site
page. Changing assignments with the word clouds and its execution.
Lesson 5. Creation educational games with triggers or hyperlinks in
MS PowerPoint.
Lesson 6. Social expertise of the games (organized with Google Drive – a
spreadsheet). Watching the movie "The image of the modern student” by Michael
Wesch (Kansas State University, 2007). Creating a collective essay "How I like to
learn" in online document on Google drive.
Lesson 7. Co-creation of mind map "Modern teacher." Taking final poll and test.
Making a poster “It’s interesting to know” and contributing to wiki page
“Communication in the network” are for independent work.
Returning to the problem of assessment, it should be noted that in presentation
"My Choice" 68 students described their school successes and achievements
considering them as the next step in their professional development, but some
students evaluate these achievements as formal, not real:
"To the moment of leaving school I had about 60 letters of honor in my
"collection". But I understand they only have formed my ability to learn
and now mean nothing".
It’s necessary to change assessment system shifting the emphasis from control to
formative function for strengthen the motivation. In addition it’s necessary to vary
teacher’s assessment with self-assessment and social assessment due to teacher’s
criteria. Students can determine the quality of the work due to criteria and assess
technical realization, structure, relevance to age of pupils, subject, design, quality of
illustrations, literacy, etc. This forced students to think about the quality of their work
while creation it.
Social expertise has stimulating, diagnostic and formational functions. A wish to
get social appreciation motivates to create high-quality bright individual works. Such
way of assessment forms critical perception of the information. A student with low
self-esteem at first is afraid that his group mates will assess not his work but his
person, but it has never happen, all students are aimed to be objective and
independent judges.
The social aspect is a priority for a new generation. They tend to have a
relationship regardless of the field of interaction. Education is not an exception. If the
teacher does not build the relationship consciously, it usually takes the worst form. A
teacher should know that a grade is no longer a sufficient motivation for the positive
attitude of the student to the discipline, especially in the pass-fail system. It is clear
that students often associate the discipline with teacher’s personality. The high quality
206 N. Kushnir, A. Manzhula and N. Valko
executed tasks show not only an attempt to get a high grade but sympathy to a
teacher. In computer disciplines social aspect is often nearly absent or poorly
developed: most of the time students work individually on the computers.
Fig. 6. Increasing students’ achievements by the method of formative assessment
The times of detached teacher have passed. Today’s successful teacher can easily
contact with the audience in a short time, acts as a partner, friend and a leader.
During the course, we found that the most third-year students have lack of
communication. For example, at first lesson after the game "Introduction", we asked
all students to tell us about their three small victories. Such "victories" were often met
with utter surprise of the group. It was uncovered that the students' ability to learn is
directly connected with their sense of self-confidence and positive mood.
Obviously, a success of training approach strongly depends on a teacher
personality and his skills in training and teambuilding. Therefore, the authors of this
paper have selected games and exercises that do not require special teacher’s skills
and knowledge or any additional devices.
Training elements in students’ groups of different teachers’ specialties and years of
study had different reaction. Some students expressed stunning and rejection: "Why
should I do that? It is not serious." despite the fact that they had finished Innovative
Teaching Methods and Technologies discipline, training exercise was an
extraordinary situation. One of the teachers said describing her experience: "Third-
year students look so serious, that sometimes I feel scared to come to the classroom,
as if teachers and students have switched their roles...”. We find it particularly
important to use the elements of training and communicative techniques working with
students of teaching specialties. Present students are future teachers who are taught
traditionally. In several years, they’ll copy one of teaching styles they have seen. It is
important to implement innovative techniques and give to students a chance to test
them in practice.
Another important improvement in the organization of the discipline was using of
open online document on GoogleDrive among teachers. Each of the teachers
contributed to planning, selection of training exercise and other notes. After each
lesson, teachers made a record in “a discipline online diary” to describe in free form
results of the lesson, such as the most common student mistakes, teaching and
New Approaches of Teaching ICT to Meet Educational Needs … 207
technical problems, positive situations, questions from the students, etc. Keeping a
discipline diary had a great success. Teaching has become more effective and
convenient, the level of awareness and collaboration among teachers has risen, so
now teachers create and use it in other disciplines.
4 Results and Discussion
Developed approaches allowed us to improve a range of disciplines including
"Information Technology", which is taught for students of all teacher specialties. We
also applied them to following disciplines: "Introduction to Information Technology"
(for future teachers of elementary school and Computer Science, the 1st year of
studiy), "Fundamentals of Computer Science and Applied Linguistics" (translators,
the 2nd year of study), and “Office Computer Technology” (programmers, the 1st
year of studies).
Students expressed their positive attitude verbally in the classroom and several
students sent e-mails with gratitude after finishing the discipline.
The results of the final poll confirmed their appreciation.
Fig. 7. Word cloud composed of students' comments about IT discipline
5 Conclusions and Outlook
Thus we have obtained the following results:
1. Discussed and analyzed characteristics of generation Net students;
2. Highlighted the contradiction between the characteristic of today's students and
traditional ways of ICT teaching;
3. The approaches to resolve these contradictions are found and implemented in
teaching practice.
In the future we plan to expand the list of pedagogical specialties to which the up-
dated version of the course "Information Technology" will be taught and improve
other disciplines according to the proposed approaches.
208 N. Kushnir, A. Manzhula and N. Valko
References
1. ICT in Higher Education in CIS and Baltic States: State-of-the-Art, Challenges and Pros-
pects for Development. Analytical Survey, 6, GUAP, St Petersburg (2009)
2. Berk, R. A. Teaching Strategies for the Net Generation. Transformative Dialogues: Teach-
ing & Learning Journal, 3(2), 1–23 (2009)
3. Cashmore, P.: Stats Confirm it: Teens don‘t Tweet. Nielsen NetView Audience Measure-
ment Survey, July 2009, http://mashable.com/2009/08/05/teens-dont-tweet (2009)
4. DeAngelo, L., Hurtado, S. H., Pryor, J. H., Kelly, K. R., Santos, J. L., Korn, W. S.: The
American College Teacher: National Norms for the 2007–2008 HERI Faculty Survey.
Higher Education Research Institute, UCLA, Los Angeles (2009)
5. Frand, J. L.: The Information Age Mindset: Changes in Students and Implications for
Higher Education. EDUCAUSE Review, 35, 15–24 (2000)
6. Greenberg, E. H., Weber, K.: Generation We: How Millennial Youth are Taking over
America and Changing our World Forever. Pachatusan, Emeryville, CA (2008)
7. Horrigan, J. B.: Home Broadband Adoption. Washington, DC: Pew Internet and American
Life Project (2006)
8. Horrigan, J. B., Rainie, L. Internet: the Mainstreaming of Online Life. Washington, DC:
Pew Internet and American Life Project (2005)
9. Junco, R., Mastrodicasa, J.: Connecting to the .Net Generation: What Higher Education
Professionals Need to Know about Today’s Students. Washington, DC: Student Affairs
Administrators in Higher Education (NASPA) (2007)
10. Kridl, B.: The condition of education. National Center for Education Statistics (NCES),
Washington, DC, U.S. Department of Education, Office of Educational Research and Im-
provement, National Center for Education Statistics (2002)
11. Ostrow, A. Stats: Facebook Traffic up 117%, Veoh Soars 346%. Nielsen Net Ratings, Au-
gust 2007, http://mashable.com/2007/09/13/nielsen-august (2009)
12. Pryor, J. H., Hurtado, S., DeAngelo, L., Sharkness, J., Romero, L. C., Korn, W. S., Tran,
S.: The American Freshman: National Norms for fall 2008. Higher Education Research In-
stitute, UCLA, Los Angeles (2009)
13. Sax, L. J., Ceja, M., Terenishi, R. T.: Technological Preparedness among Entering Fresh-
man: the Role of Race, Class, and Gender. Journal of Educational Computing Research,
24(4), 363–383 (2001)
14. Tapscott, D.: Growing up Digital: How the Net Generation is Changing your World.
McGraw-Hill, NY, (2009)
15. Behrstok, E., Clifford, M.: Leading Gen Y Teachers: Emerging Strategies for School
Leaders. TQ Research&Policy BRIEF, Washington, DC, USA (2009)
16. Petuhova, L.E.: Theoretical and Methodological Background of Informational competen-
cies Formation of Elementary School Teachers. Doctoral Dissertation, Specialty 13.00.04 -
Theory and Methods of Professional Education. The South Ukrainian National Pedagogi-
cal University named after K.D. Ushynsky, Odesa (2009)
17. Kushnir, N., Manzhula, A.: Formation of Digital Competence of Future Teachers of Ele-
mentary School. In: Ermolayev, V. et al. (eds.) ICT in Education, Research, and Industrial
Applications. Revised Extended Papers of ICTERI 2012, CCIS 347, рp. 230-243, Springer
Verlag, Berlin Heidelberg (2013)
Pedagogical Diagnostics with Use of Computer
Technologies
Lyudmyla Bilousova1, Oleksandr Kolgatin1 and Larisa Kolgatina1
1
Kharkiv National Pedagogical University named after G.S.Skovoroda, Kharkiv, Ukraine
lib215@list.ru, kolgatin@ukr.net, larakl@ukr.net
Abstract. The technology of the automated pedagogical diagnostics is ana-
lysed. The testing strategy, oriented for pedagogical diagnostics purpose, and
grading algorithm, which corresponds to Ukrainian school grading standards,
are suggested. "Expert 3.05”software for automated pedagogical testing is de-
signed. The methods of administration of the database of the test items are pro-
posed. Some tests on the mathematical topics are prepared with "Expert 3.05".
The approbation of these tests in the educational process of Kharkov National
Pedagogical University named after G.S.Skovoroda is analysed.
Keywords. E-learning, Diagnostics, Test
Key terms. InformationCommunicationTechnology, Teaching Process
1 Introduction
Pedagogical diagnostics is the integral part of adaptive E-learning courses. The un-
conditional quality of testing is its high informative abilities. However, in practice the
large part of the test information is often not used. Computer technologies give us
possibility to organize the qualitative pedagogical diagnostics at new level. Modern
automated systems which can be qualified as expert systems are capable to supply
comprehensive algorithms of testing and analysis of the test results. Testing with use
of computers allows a teacher to obtain the summary characteristics of knowledge and
skills of the pupils' group and to use this information to choose the teaching methods.
A study of such algorithms is a wide field of the scientific work. Therefore, the aim of
our paper is to design methods of the pedagogical diagnostics, which satisfy following
demands:
Different forms of the intellectual activities of an examinee are attracted in process
of testing;
The automated system of the pedagogical diagnostics ensures its diagnostic abili-
ties at wide differences of the examinees mastering;
Processing of the test results provides maximum information for an examinee and a
teacher to correct the educational process.
210 L. Bilousova, O. Kolgatin and L. Kolgatina
2 Objectives
The first stage of pedagogical diagnostics organising is a construction of an idealised
pedagogical model that is allocation of basic elements of knowledge and skills, as
well as detecting the level of its mastering.
The second stage represents creation of the problems system which covers all ele-
ments of knowledge and skills and all levels of their mastering.
We cannot design test as a system of test items of equal difficulty, in spite of rec-
ommendation of the classic test theory. Such approach gave the best tests for dis-
crimination of examinees into several groups. However, the test with equal items has
low validity for examinees with bad mastering because of guessing answers. Validity
of such a test is also low for good mastering examinees because of lack of attention.
Therefore, it is certainly necessary to include problems of different difficulty to the
test.
How to design a test item of advanced difficulty? What is difficulty? Why the most
of examinees do not solve some problems?
We cannot use problems of the reproductive level as items of advanced difficulty.
There are not difficult facts and easy facts. Our educational process should be organ-
ised to provide steady knowledge of all compulsory facts. If the most of examinees do
not know some compulsory facts, it means that we should correct our teaching. We
are against using items which correspond to facts that are fragmentary studied and
which are not basic for the tested topic. Therefore, all problems corresponding to the
reproductive knowledge must have equal difficulty.
Someone can increase the difficulty of an item by combining several operations in
this item. Such approach leads to increasing influence of lack of attention on the test
results, as well as to necessity of using weight coefficients and to decreasing of the
measuring accuracy of the test. We are also against using problems which correspond
to facts that are fragmentary studied and do not form basis of the topic being tested.
In our opinion, the item of advanced difficulty should be connected with use of
more difficult, not reproductive kinds of the intellectual activities [1], [2].
Full and high-qualitative pedagogical diagnostics should be built on the system of
test items of all levels: reproductive and productive. By analogy with levels of educa-
tional achievements [2], which are standardised by the Ukrainian Ministry of Educa-
tion and Science [3] we propose the following levels of the test items:
1. Initial level - it is the very simple problems which assume the reproductive charac-
ter of the student’s activities, mainly distinguishing. The difficulty index of these
test items is about 1, the most of the examined students execute these items cor-
rectly.
2. Average level - it is the problems which assume the reproductive activities, these
problems cover all basic facts and unary skills according to curriculum. A database
of items of this level is designed the most naturally. According to the Ukrainian
standards [3], the student can continue education, if he (she) knows not less than
50% of compulsory facts determined by curriculum. Therefore, by linear estima-
tion, average index of difficulty must be near 75% for the reproductive items.
Pedagogical Diagnostics with Use of Computer Technologies 211
3. Sufficient level - these items assume the examinee applies his knowledge and skills
for solving problems in standard situation.
4. High level - these test items are practical problems which assume executing of new
algorithm, carrying knowledge into new, non-standard situation, etc. These items
can lose creative nature, if the method of its solving was explained in the process
of learning. Therefore, the database of the items of the level 4 requires continuous
analysis and modernisation.
We propose the vector processing of the test results – separate calculation of the score
for items of every level. It allows to avoid the use of the artificial weight coefficients
and to provide the comprehensive algorithm of adaptive strategy of testing and grad-
ing. We also propose the separate processing of the results for the test items according
the elements of knowledge and skills.
Using computer for test administration allows to analyse the examinee's results di-
rectly in process of testing and to suggest an examinee the items, which mostly corre-
spond to his (her) level of educational achievements. Such approach is often called
adaptive or quasi-adaptive testing [4].
3 Model and Algorithm
A choice of the items level for start of testing is an important question of the adaptive
strategy. Testing usually starts from the simplest items. Such approach provides de-
creasing of psychological discomfort and creates the atmosphere of competition, the
feeling of growth according to complication of the problems. Taking into account this
consideration we propose to start testing just from items of the level 2 which are the
simplest for the examinees that will obtain positive grade.
There is additional argument to choose the level 2 as the start level of testing. The
test items of the level 2 reflect the compulsory facts of the study topic; these problems
cannot be excluded from the test process. It is not worthwhile to start test from the
items of the level 3, because productive and, especially, creative problems are based
on sufficiently wide spectrum of knowledge, and it is not always possible to detect,
which exactly element of curriculum is not mastered by an examinee. The items of the
level 1 are intended for students whose mastering is not satisfactory; therefore, there
is no need to suggest these items to all examinees. Our testing strategy and algorithm
of grading are presented on the fig. 1.
Here are some comments to fig. 1. The testing starts with the items of the level 2.
An examinee solves the compulsory minimum of the items on the level 2, automated
system calculates S2 - his (her) score on the level 2 and estimates the error of the
score. If accuracy is enough, the automated system chooses a grade or increases the
level of items, which are being suggested to the examinee. Otherwise, the items of the
level 2 are being suggested to the examinee until the accuracy become satisfactory. It
should be underlined that accuracy depends not only on the number of items, but it
depends on the individual test score [5]. The necessary accuracy is also conditioned
212 L. Bilousova, O. Kolgatin and L. Kolgatina
by the differences between the test score and the key points for decision about grading
or a rise of a level.
Begin
Solving of problems of the level
2
N Yes
S2>=0.5
Solving of
problems N S2>=0.8 Ye
of the
S1>=0.8 Yes
N
Grade 3
N Ye
S1>=0.5
А
Grade 1 Grade 2
А А
N Ye Solving of
S2>=0.6 problems
Grade 5 of the
N S2>=0.5 Ye А
N S3>=0.8 Ye
Grade 3 Grade 4
Solving of
А А problems
of the
N S4>=0.7 Ye
N Ye N Ye
S3>=0.5 S4>=0.5 Grade12
Grade 8
N Ye Grade11 А
N Ye S4>=0.3
S3>=0.3
А
А
Grade 6 Grade 7 Grade 9 Grade10 А
А А А А
End
Fig. 2. Grading algorithm
Testing on the levels 3, 4 and 1 is carried out by analogy with level 2 with scores
S3, S4 and S1 accordingly.
Such algorithm of testing improves accordance between the level of items being
suggested and the level of the examinee's mastering. Influence of a lack of attention
on the test score for examinees with an excellent mastering is decreased. The exami-
Pedagogical Diagnostics with Use of Computer Technologies 213
nees with bad mastering solve easy problems, which correspond to the most important
parts of the educational content. The psychological discomfort, connected with con-
stantly incorrect answers, is excluded, but such easy items give a possibility to deter-
mine the structure of the knowledge and skills, as well as to distinguish examinees,
who have not mastered the compulsory minimum according to curriculum.
In any case the result of the pedagogical diagnostics will be more careful in com-
parison to testing without special selection of items.
4 Software
The computer support of offered technology is provided with information system
"Expert 3.05" designed by us as a distributed database in the MicrosoftAccess2003
environment. The important advantage of our information system in our opinion is a
modular principle of its construction which allows the author of the test items (proba-
bly, with the help of the programmer) to create and to add to the database new forms
of the test items. The central database of information system is the test items database.
The test items are grouped by topics for convenience of viewing. The elements of an
educational content are picked out in each topic. To each element the author provides
the comment for a student who has not mastered this content. Some blocks of the test
items of a different level of difficulty are offered to verify student's mastering in each
element of an educational content. The author specifies for the every block such pa-
rameters: a test items level (0-4), a weight factor and maximum time of exposition of
one item. All items of the block should be of one type, that is, the identical dialogue
form. The student will be offered one or several items from each block by a casual
choice in a process of testing. Quantity of items blocks and filling these blocks are
determined by required quality of diagnostics.
The database, which contains the information on the answers of each examinee on
each item, is formed by results of testing. This database includes such fields: the code
of the test item, the correctness of the given answer, the level of the item, probability
of casual guessing of a correct answer, time of item solving. The additional service
information (time and date of testing, examinee's grade etc.) is also being stored.
The examinee receives (fig. 2) the diagnostic data on each element of knowledge;
chart, which reconstructs a structure of his (her) knowledge; recommendations for
independent work. The author receives the statistical analysis of each item difficulty
and discrimination, correlation with the test score (grade). The diagram shows de-
pendence of item difficulty on the examinee's test score (grade) and is very useful
(fig. 3). The author has an opportunity to generate database query by means of Access
environment and to pass the data for the further analysis in spreadsheets.
214 L. Bilousova, O. Kolgatin and L. Kolgatina
Fig. 3. Diagnostic data for the examinee
Fig. 4. Statistical analysis of a test item for the author
5 Experience of Diagnostics with “Expert 3.05”
We have prepared the system of the test items with "Expert 3.05" on some courses:
“Mathematical methods in psychology”; “Theoretical basis of informatics”; “Archi-
tecture of a personal computer”.
Now we are able to start the third stage of the preparation of the system of the
pedagogical diagnostics - approbation and verification of the test items.
Pedagogical Diagnostics with Use of Computer Technologies 215
Our technology of verification is based on the requirements of the Standard of the
Ukrainian Ministry of the education and science [6] and takes into account features of
the automated pedagogical diagnostics.
The analysis of the test begins with detecting of a level of educational achieve-
ments of the students based on the expert's rating, for example, it may be a traditional
examination. The complete verification procedure assumes that the experts determine
such rating irrespective of the verifying test. Approbation should be organised with
enough number of examinees to guarantee sufficient number of answers for any test
item. Determination of the student's rating by experts cannot be organised so often, as
it is necessary for constant updating of the test items database. Therefore, for the cur-
rent verification it is possible to offer detecting of a level of student’s educational
achievements with the help of the same automated system of pedagogical testing and
to check a correlation of the separate item with the test score. In such a case, the ap-
probation data are accumulated continuously, including the independent work of the
students with the automated system. Validity of the automated system of testing as a
whole is checked through comparison of integrated results of testing with results of
other kinds of the control: interview, examination, execution of practical works etc.
For maintenance of reliability of the current verification, the automated system does
not include into the analysis the answers of not registered examinees: teachers and
other users, whose names are not in the lists of the students groups. The answers of
the test pass, which has not been finished, are not analysed too. The answers are not
taken into account, if the time of its execution is smaller, than it is necessary for ac-
quaintance with the text of the problem. There is an opportunity to specify additional
conditions of selection of the valid answers with use of the Access environmental (for
example, date and time of testing, educational group, variant of the test etc.).
After distribution of the students according to their educational achievements, the
conformity of a level of the test item and its empirical index of difficulty is checked.
According to the requirements of the Ukrainian educational standards, the students
with an average level of educational achievement (the grade of 4 on the grading scale
with 12 grades) "... knows about half of educational content, is capable to reproduce
it, to repeat after the model the certain operation..." [3]. Just the items of the level 2 in
our classification have such contents. Thus, the difficulty index of the items of the
level 2 can not be less than 0.5 for such students and we consider it to be within the
range 0.5-0.9. It is necessary to note that such a range of difficulty is not convenient
from the viewpoint of improvement of test statistical parameters. However, the items
of the level 2 represent a set of educational content facts, which are obligatory for
mastering. Therefore, the problems author cannot change its difficulty without chang-
ing the curriculum.
Thus, for the items of the level 2 on the sample of students with an average level of
educational achievements we have the following algorithm of analysis of the item
quality:
An item does not require correction, if its difficulty index within 0.5-0.9, the dis-
crimination index is higher than 0.25, that is, discrimination index satisfies the re-
quirements of the standard [6] (fig. 4.).
216 L. Bilousova, O. Kolgatin and L. Kolgatina
The item difficulty index is more than 0.9. This item should be analysed with the
help of the diagram of dependence of difficulty from educational achievements of
the students. It should be determined, whether this item has the discrimination abil-
ity for the students with the initial level of educational achievements, accordingly,
this item level should be changed (fig. 5). Otherwise, this item should be removed
from the test.
The difficulty index is less than 0.5, it is necessary to analyse the content of the
item, such situations are possible:
─ The item is not reproductive, its discrimination ability satisfies the requirements
of the standard. It is necessary to increase the item level (fig. 6).
─ The item has low discrimination ability for all students; it signifies that some
mistakes in the formulation of the item take place. This item should be removed
or corrected (fig. 7). There is a possibility of the situation, when all experts
agree that the problem is correctly designed and satisfies the curriculum, in such
a case mastering of the students should be checked by some another method
and, may be, the quality of educational process should be analysed.
The best range of difficulty index for items of levels 1, 3, 4 will be within 0.5-0.6
(allowable 0.3-0.7) on the sample of the examinees of appropriate level of mastering.
The analysis of these items on a difficulty is carried out by analogy with level 2.
Fig. 5. A typical item of the level 2
Pedagogical Diagnostics with Use of Computer Technologies 217
Fig. 6. A typical item of the level 1
Fig. 7. A typical item of the level 3. It is not a reproductive problem, if the students do not learn
the table of the binary codes for numbers until 15.
After finishing three stages of preparation the system of pedagogical diagnostics is
ready for practical use. The stage of practical application of the system combines
procedures of testing and statistical processing of the obtained results including the
interpretation of the results for students, teachers and authors of the test problems.
The expert system of pedagogical diagnostics needs continuous modernisation of its
database. Naturally, it requires returning to previous stages of the work with the sys-
tem.
218 L. Bilousova, O. Kolgatin and L. Kolgatina
Fig. 8. An unsuccessful item
The “Expert 1.01-3.05” software is used in Kharkiv National Pedagogical Univer-
sity named after G.S.Skovoroda since 2001. Here are some latest results on approba-
tion. The test on mathematical methods of statistical analysis of the pedagogical diag-
nostics data was suggested to future teachers of informatics, mathematics, chemistry
as an element of courseware "Information systems in pedagogical activities” in 2012-
2013 academic year. The purpose of testing was the self-diagnostics of students. So,
the students could pass the test many times, studying the problem elements of the
learning content and improving their results. We took into account the best results of
testing and compared its with the examination results. The Pearson correlation was
0.7 at sample of 51 students and we can consider it as the test validity.
The test results gave us possibility to study the structure of students’ knowledge at
basic questions of statistical analysis of the pedagogical diagnostics data (fig. 8). The
error estimation for the data on fig. 8 was evaluated as a half of the 95% confidence
interval
1.95s ,
y
n
where s is the estimation of the standard deviation and n is the number of test items of
the given learning element, which were passed by students. The errors are different
for every point on fig. 8, because the number of test items on various elements is dif-
ferent, so we show the ranges of errors in table 1.
The results (fig. 8) show that problems of choosing the scales for the pedagogical
evaluations are the most difficult for students. The problems of reproductive level,
where student should to choose the method or formula for estimation of some parame-
ters of statistical distribution, are the easiest. But on the productive level, when stu-
dents should explain the influence of the values and number of variants in a sample on
the estimated parameters, such problems are the most difficult.
Pedagogical Diagnostics with Use of Computer Technologies 219
Fig. 8. Difficulty index (probability of correct answer) as a function of the student’s grade for
problems of various elements of the learning content
220 L. Bilousova, O. Kolgatin and L. Kolgatina
Table 5. Errors of difficulty index estimation at different student’s grades
Student’s grade Estimation error
1 0.01-0.04
2 0.04-0.1
3 0.02-0.05
4-5 0.01-0.07
6-7 0.02-0.16
8 0.03-0.18
9 0-0.4
10-12 0-0.6
6 Conclusions
1. New comprehensive algorithm of testing and grading is suggested. This algorithm
takes into account possibilities of the computer technologies and requirements of
Ukrainian standards.
2. The automated system of the pedagogical diagnostics is designed.
3. The methods of the items database administrating is proposed and used in practice
in the educational process of the Kharkov National Pedagogical University.
References
1. Bloom, B.S.: Taxonomy of Educational Objectives. Book 1: Cognitive Domain. Longman,
Inc., New York (1956)
2. Bespalko, V.P.: Basis of the Pedagogical Systems Theory: Problems and Methods of Psy-
chological and Pedagogical Providing of Technical Teaching Systems, Voronezh Univer-
sity Press, Voronezh (1977)
3. Criterions of Grading of the Educational Achievements of Pupils in the System of Secon-
dary Education. Education of Ukraine, 6 (2001)
4. Zaitseva, L.V.; Prokofyeva, N.O.: Models and Methods of Adaptive Knowledge Diagnos-
tics. Educational Technology & Society, 7, http://ifets.ieee.org/russian/depository/v7_i4/
pdf/1.pdf (2003)
5. Kolgatin, O. G.: The Statistical Analysis of the Test with Different Forms of Items. Means
of Teaching and Research Work, 20, KhSPU, Kharkiv (2003)
6. Means of Diagnostics of a Level of Educational and Professional Training. The Tests of
the Objective Assessment of a Level of Educational and Professional Training. Order of
the Ministry of Education and Science of Ukraine, № 285, 31 July 1998 (1998)
The Use of Distributed Version Control Systems
in Advanced Programming Courses
Michael Cochez, Ville Isomöttönen, Ville Tirronen and Jonne Itkonen
Department of Mathematical Information Technology
University of Jyväskylä
P.O. Box 35 (Agora), 40014, Jyväskylä, Finland
{michael.cochez, ville.isomottonen, ville.tirronen, jonne.itkonen}@jyu.fi
Abstract. Version Control Systems are essential tools in software devel-
opment. Educational institutions offering education to future computer
scientists should embed the use of such systems in their curricula in or-
der to prepare the student for real life situations. The use of a version
control system also has several potential benefits for the teacher. The
teacher might, for instance, use the tool to monitor students’ progress
and to give feedback efficiently. This study analyzes how students used
the distributed version control system Git in advanced programming re-
lated courses. We also have data from a second year course, which enables
us to compare between introductory level and master’s level students. We
found out that students do not use the system in an optimal way; they do
not commit changes often enough and regard the version control system
as file storage. They also often write commitmessages which are mean-
ingless. Further, it seems that in group work settings there is usually one
dominant user of the system.
Keywords. Programming Education, Version Control System, Git
Key terms. ICTTool, TeachingProcess, ICTEnvironment, Technology
Introduction
Version control systems (VCSs) have a decades-long history in professional soft-
ware engineering with early systems like Source Code Control System (SCCS)
and Revision Control System (RCS) developed in the seventies and eighties re-
spectively. These pioneering systems only supported storage of the versions on
the file system, while later systems also allowed for remote and mostly centralized
storage of the versions. The most well-known centralized systems are Concurrent
Versions System (CVS) and Subversion (SVN). Currently, there is a trend to-
wards the use of distributed version control systems (DVCS) where each user has
a local copy of the repository which can be synchronized with other repositories.
Systems such as Git and Mercurial exemplify this type of present-day decen-
tralized technology. These DVCSs enable flexible change tracking, reversibility,
222 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
and manageable collaborative work, which are valuable for both small and large
projects.
There are many arguments for incorporating VCSs into an educational set-
ting. From a teacher’s point of view, using VCSs increases the possibility of
monitoring how students make progress with their assignments and eases the
feedback process. The teacher could, for instance, include corrections and sug-
gestions directly into the students’ program code [1]. More generally, educators
acknowledge that the use of VCSs relates to effective team work and that it is
a crucial skill to be taught to prepare a competent workforce for present-day
distributed workplaces [2].
An educational concern of interest to us is how students actually use VCSs.
This has been previously studied by Mierle et al. [3] who investigated VCS
usage patterns in a second-year course, hoping to find a correlation between an
effective use of VCSs and study success. No clear patterns could be identified in
the data which the authors attributed to the fact that beginner students climb
their learning curve at different rates; see [3, 4]. These authors call for more
research on VCS usage patterns in particular in upper-year courses [4], which
motivates the present study.
We have collected data about students’ use of the distributed version control
system (Git) from three different courses: Introduction to Software Engineering
(second-year bachelor), Functional Programming (master’s level), and Service
oriented architectures and cloud computing for developers (master’s level). A
hypothesis arising from teacher observations during these courses is that students
use VCS principally as a submission system rather than what it is intended to
be. By this we mean that
– students commit at the end of the class sessions or right before the deadline,
or there is only one commit per week/task,
– only one group member commits everything,
– students do not consider what file types to commit,
– overall, with no specific training, student do not use VCS efficiently.
We study these issues quantitatively exploring version control commit fre-
quencies, commit sizes and the activity of individual students. Our specific re-
search interest is the potential usage patterns identifiable in the commit log data
of Git repositories.
1 Version control systems in education
Clifton et al. [5] summarize that in educational settings VCSs have been adopted
to enable more realistic software development experiences for students [6], as a
tool to monitor or visualize team and individual contributions [7], and for non-
code artifacts such as creative writing [8]. Clifton et al. themselves, as well as
many others, use a VCS for course management purposes. Further, some authors
regard VCSs as a valuable tool to monitor and understand how students develop
code [9]. Unsurprisingly, one of the most usual educational targets appears to
The Use of Distributed Version Control Systems ... 223
be courses with project work where VCSs both foster team work and facilitate
course management tasks such as assessment and grading [10]. Milentijevic et
al. [11] go as far as to propose a generalized model for the adoption of VCSs
as support in a variety of project-based learning scenarios. All in all, we find
that there is a general consensus of the benefits of VCSs as an integral part
of computing curricula, one key argument being that they measure up to the
requirements of globally distributed workplaces [2].
There are also challenges in the educational use of VCSs. Reid and Wilson [4],
who used the CVS system, report on the confusion in judging which of the
students’ assignment versions was the final one. Glassy [9] found that students
tend to put off working on assignments for as long as possible, even though a
VCS is proposed to them with the hope of iterative work processes. Issues of
this kind relate to inefficient use of VCSs. Furthermore, Reid and Wilson [4]
noticed that some students mixed the functionalities of the CVS check out and
update commands, and that also teaching assistants encountered problems if
they had not properly familiarized themselves with the tool. These issues were
considered to be due to a lack of a mental model of the VCS system used.
Yet another challenge Reid and Wilson [4] raise is increased teacher workload
when repositories are initiated and managed by teachers. In a more recent study,
Xu [10] points out that there can be a long and rough learning curve before
students feel comfortable using Git. Accordingly, Milentijevic et al. [11], who used
CVS, report that students find a VCS to be a useful tool after they are sufficiently
familiar with it. In the paper by Glassy [9] and Xu [10], the value of informative
commit log messages is raised as a topic to be emphasized to the students.
Rocco and Lloyd [12] in turn observed that some student have difficulties in
understanding what constitutes “a significant change” to be committed.
It is much more difficult to find systematic empirical studies on issues such
as how frequently students make commits and how they share the work. Rocco
and Lloyd [12] found in their data that over 80% of a CS1 course population
could adopt an iterative work process with the Mercurial system (50.0% did 7–
21 commits and 33.3% more than 21 commits). On another course the authors
defined a minimum commit frequency for one assignment and no requirements
for the assignment that followed. With the first assignment, 75% of the students
obtained a reasonable commit frequency, while with the latter this was 81%,
altogether indicating that informing students of proper VCS usage can have a
positive effect on their work processes. The authors note that not only were the
students able to grasp the basics of the VCS (Mercurial), but they tended to
continue to take advantage of the tool later on.
The present study focusing on the students’ usage patterns with the Git
system in both a second-year course and master’s level courses complements
the studies such as the ones by Rocco et al. [12] and Mierle et al. & Reid and
Wilson [3, 4].
224 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
2 The courses
Introduction to Software Engineering (SE) is a 3-credit course consisting of lec-
tures, a course assignment, and an end-of-course exam. The lectures introduce
students to the basic concepts of software engineering, while the mandatory
course assignment is the preparation of a project plan. The assignment is done
in small groups and consists of four larger phases that need to be accepted by
the lecturer. Mandatory supervision sessions on version control were arranged
at the beginning of the course in order to encourage all the students to use the
distributed version control system Git for the group assignment. The course had
altogether 72 students in 33 groups (2.18 ± 0.76 students per group).
Functional programming (FP) is a 6-credit course implemented without tra-
ditional lectures and exams. The course is run in week-long cycles such that
each week a new set of exercises is announced for the students. Students work in
small groups and all of their study time is devoted to programming the weekly
exercises. Two contact sessions are held each week. The first one is devoted to
supporting the students’ work and answering their questions. During the second
weekly contact session there is a review of the student-written code. Overall,
the course emphasizes self-direction on the part of students, similar to recently
discussed course models such as the flipped classroom; see more details in [13–
15]. Git was proposed for students as their primary group work tool and all of
the exercises had to be returned via it. Thirty-six students where active in the
course divided over 13 groups. (2.77 ± 0.89 students per group)
The last master’s level courses studied, Service oriented architectures and
cloud computing for developers (SOA&CC ), introduces students to the use of
digital services and the concept of cloud computing. A format similar to the FP
course is used during the first (5 credits) part of the course. During that part
of the course students undertake independent group work on a set of assign-
ments each week. Two weekly sessions are arranged for the group work and one
mandatory contact session focusing on reflective program review is arranged at
the end of each week. An analysis of how the course model used in this course
attempts to motivate students can be found in [16]. During the course Git is not
only used as a version control system; it is also used as a tool to deploy code
to Platform as a Service (PaaS) providers. Nine groups of students were formed
with altogether 36 students (4 ± 0.82 students per group).
All three courses utilize the Faculty’s YouSource 1 system. Similar to staff
members, students can use their university credentials to log in to this system and
create projects and Git repositories to manage collaborative work. The projects
and repositories can be defined to be either private or public and collaborators
can be added to them with a variety of permissions. This system has been in
use at the department since mid-2010 and has been used in many courses and
research projects.
It should be noted that in the remainder of this paper we are specifically
concerned with the Git version control system, which belongs to the third gen-
1
https://yousource.it.jyu.fi/
The Use of Distributed Version Control Systems ... 225
eration of version control systems (DVCSs). Students are free in their choice of
environment for interacting with the version control system. Students can for
instance use the git command line tool, tools with a graphical user interface, or
tools included in their integrated development environments.
3 Data analysis
The Git repositories which students or course teachers created for the respective
courses on the above-mentioned YouSource system are the source of all data
analysis in this paper. One limitation of this data source is that we cannot see
any data related to branches which a student did not push to the central Git
server. However, if work of one of these so called local branches got merged
into a branch which is synchronized with the central Git server, we are able to
see its history as well. Further, this limitation is of minor importance since we
are mainly interested in how students use the version control system in group
work settings. Another limitation, which is inherent to the Git DVCS, is that
we cannot know for sure whether time stamps on commits are truthful. It is
technically possible to tamper with the date of the commits, but since there is
no benefit for students to do so, we make the assumption that the time stamps
are correct.
To study our research hypothesis, we will perform five different analyses, the
first four of which are based on commits to the repository and the last one on
the content of the repositories. For each commit we extracted the number of
insertions and number of additions in accordance with the short status log of
each commit2 . We added these two numbers together to form what we will call
the number of changes of that commit. The tools used in the analysis have been
developed by the authors of the paper and consume output produced by the
diverse git commands.
For the first analysis we will, for each course, look at the commit activity over
the whole course. To be concrete, we will visualize the commit activity by plotting
the estimated probability density function of the total number of changes, i.e.
for all students, over the span of the course. The density is estimated via the
standard kernel density estimator, using a Gaussian kernel with bandwidth of 6
hours.[17] The height of the plot then shows the relative likelihood of a commit
at a specific point in time.
The second analysis focuses on students’ activity during the implementation
sessions. This is done only for the FP and SOA&CC courses since the SE course
does not have distinct sessions during which students get time to implement their
work. We use a similar method as in the first part, but accumulate all commits
that were made during the implementation sessions in the same plot. This plot
shows when the students commit their code during the contact sessions. In the
figure the far left of the x-axis represents the start of the session and the far
right 15 minutes after the end. This is done in order to account for commits
2
http://www.kernel.org/pub/software/scm/git/docs/git-log.html
226 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
right after the sessions. In this case we use a bandwidth which is one tenth of
the total length of the session.
In the third part we perform an analysis of the commit messages in the
different courses by classifying them in three categories : useful, trivial, and
nonsense. A message is placed in the nonsense category if its content is not
anyhow related to what is being committed. An example of this type of messages
are these which contain only a couple of random letters, needed because the git
system does not allow for empty commit messages. A trivial message is one
which has no information beyond what is immediately visible from the commit
meta-data. This category includes, for instance, a message consisting of a list
of changed files or one saying that a given commit is a merge of two branches.
All other commits are classified as useful. It should be noted that being in the
useful class does not directly imply that the message is of high quality. It only
means that the message is not trivial or nonsense. The classification was done
manually by the respective teachers of the courses. We do not try to make a
comparison between the courses, because the bias caused by having different
raters is difficult to estimate.
In the fourth part we try to measure whether the version control system is
used equally among the students in the group. If the system would be used by
all students in a group, we would expect that the most active student in a group
of n students performs (1/n) ∗ 100% of the commits. To represent this number
for all groups in the different courses we first find the students with the highest
number of commits in their respective groups. Then we calculate their individual
share in the total number of commits of their group. We then create an overview
of the obtained percentages where we show different graphs for different group
sizes since comparison among unequal group sizes would lead to biased results.
It only makes sense to measure this for groups with more than one person. The
SE course had a few single-person groups, hence only 28 groups from that course
are included.
For the fifth and last part we investigate the types of files which students
put under version control. First, teachers of each of the courses listed the file
types and limits which they would expect a normal repository to contain. We
started out from the files included in the HEAD of the master branch. For the
FP and SOA&CC course we determined the type of each file using the BSD file 3
command. The SE repositories required a manual analysis to decide the type of
the files because the file command is unable to distinguish between the file types
in use in the course. Then we counted the number of files of each file type. Then
for each count, we compared it to the number of files expected by the teacher
and any surplus was counted as garbage. The final number which we calculated
for each group is the fraction of garbage in the total number of files.
3
http://www.openbsd.org/cgi-bin/man.cgi?query=file
The Use of Distributed Version Control Systems ... 227
4 Results
This section describes the results of our analyses. The first subsection shows
the results for the analysis of the student activity during the whole course.
In the second subsection, we focus on the implementation sessions only. The
results of the analysis of the commit messages is shown in subsection three. Then
we consider the activity distribution among students in the fourth subsection.
Lastly, we look at the types of files which students submit to the version control
system.
4.1 Commit activity over the whole course
The student activity in the SE, FP and SOA&CC courses is presented in the
Figures 1, 2, and 3, respectively. In the SE course, we draw thick vertical lines
to indicate the end of each of the four phases of the course assignment. As can
be seen in that figure, there seems to be no correspondence between these dead-
lines and the student activity. In this course where VCS training was provided,
students appeared to commit rather evenly throughout the course. We attribute
the activity spike at the start of this course to the students trying out and get-
ting familiar with the version control system at the point in time of the training
sessions.
In the graphs of the FP and the SOA&CC courses (figure 2 and 3), we
indicated with thin vertical lines the sessions during which students get time
in the classroom to work on the assignments. The thicker vertical lines indicate
deadlines for the weekly assignments. The dates of the sessions are displayed on
the x-axis in a month/day format. In contrast to the SE course, these graphs show
a closer correlation between student activity and the implementation sessions and
the deadlines. The graphs suggest that most of the work was committed during
the contact sessions, which again suggests that students bring their work to the
sessions to be committed there. This prompts us to study the student behaviour
during the sessions separately in the next section. It is also clearly visible that
the students have a very low activity during the weekends.
4.2 Commit activity during the sessions
In the FP and SOA&CC courses students were more active during sessions than
at other times. The graphs in figures 4 and 5 show the students activity during
the sessions and 15 minutes after the session. We normalized the duration of
the session (90 minutes) and the 15 minutes overtime between zero and one.
Interestingly, we notice a similar behavior in both courses. There seem to be
three periods of higher activity. The first moment of higher activity is in the
beginning of the session after about 10 minutes. The second one, which last
longer, is between 20 and 40 minutes after the start of the session and lastly, the
activity peaks shortly after the session.
The first period of activity is most likely because individual students have
been implementing parts at home. These students then decide to commit only
228 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
Fig. 1. Commit activity during the SE course
Fig. 2. Commit activity during the FP course
Fig. 3. Commit activity during the SOA&CC course
The Use of Distributed Version Control Systems ... 229
after receiving consent from other group members. This is an indication that
students do not know how to use the version control system efficiently, as in
principle they could have used a separate branch for their local development and
merged their changes to their shared version of the exercises. Also, speculating
based on student dialogue, some students might have feared ’losing face’ by
making their preliminary versions visible to others, including the teacher.
During the second period of activity students are using the system as they
are supposed to, committing changes regularly. Then the activity drops for quite
some time before reviving shortly after the session. We attribute this last peak
to those groups who have been working during the whole session without com-
mitting many changes. At the end of the session they want to store their work
for later continuation and decide to put all their work in the system.
Fig. 4. Commit activity during the sessions of the FP course
Fig. 5. Commit activity during the sessions of the SOA&CC course
230 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
4.3 Commit message analysis
The classification of the commit messages was performed for the SOA&CC and
FP courses and yielded the results shown in table 2.
Table 1. Categorization of the commit messages per course
useful trivial nonsense
SE 996 (67%) 430 (29%) 59 (4%)
FP 1422 (78%) 276 (15%) 129 (7%)
SOA&CC 289 (74%) 65 (17%) 37 (9%)
Table 2. Categorization of the commit messages per course
In an ideal repository we would not find any trivial and nonsense messages.
What we see from the table however is that there is a significant amount of these
types of messages.
It is not visible from the table, but the teachers classifying the messages
shared the opinion that the messages in the useful category where not all that
descriptive. Some commit messages could be regarded as ‘locally sensible’, mean-
ing that they could be useful for communication during a short time span, but
offer not much for later inspection. Many of the commit messages are clear in-
dicators that the students regard the system as an answer submission system.
Examples include “Answer for week 12” and “exercise 4a”. We also noticed some
messages related to problems in using the git system. The amount was however
not as significant as the teacher had expected. It is also observed that the quality
of the messages is depending on the group, indicating that some groups use the
messages for communicating, while others do not.
4.4 Differences in student activity
To show the differences in activity among students we assembled the charts in
figure 6. The figure contains one pie chart for each course and each group size
or none if there are no groups of the given size in the course. Each pie chart
illustrates the fraction of the groups which have a given percentage of commits
for their most active committer. The last column shows the expected fraction,
i.e. the chart which would be obtained if all students in the groups do an equal
number of commits.
What we see from the charts is that the most active committer in a group,
most of the times, commits significantly more as the expected percentage. Put
another way, the most active committer in each group is very often much more
active as the average which one might expect. This can be due to that student
having a dominating role in the group. In the FP and SOA&CC course we tried
to mitigate this effect by grouping students according to their skill level.[13, 15,
16] We however think that the main differences are caused by a different level
The Use of Distributed Version Control Systems ... 231
of familiarity with the version control system between the group members. The
person with the most experience will commit more frequently or is given the
task of submitting the work of others to the system.
4.5 Which files did the students commit to VCS?
The presence of files which do not belong in the version control system, such
as executable programs, temporary compilation files and copied documentation,
suggests that the students used the VCS as plain file storage. Figure 7 shows the
fraction of the groups with a given percentage of redundant files in their final
repository. We see a big difference between the figures for the respective courses.
In the graph for the SE course we see that most groups did not include many
compiled files in their repositories, as was explicitly instructed in the course.
During the functional programming course, students can often test their code
without actually compiling it, which could explain the low amount of garbage
in the repositories. The garbage that is committed consists entirely of compiled
binaries and other compiler generated files.
During the SOA&CC course many students use integrated development en-
vironments (IDE) which do the compilation automatically for the user. It seems
like many students have included all files which the IDE produced to the version
control system.
It seems that if students are not made aware of the fact that they should
not include this kind of files to the VCS, they tend to include everything that
happens to be present in their local directory. We should do further research
to see whether this behavior changes if students are made aware of the best
practices.
Conclusion
In this article, we focused on students’ usage patterns in advanced courses re-
lated to programming while using the distributed version control system Git. We
first looked at when students commit their work during the course and in more
detail at their committing pattern during the implementation sessions. We con-
cluded that most students commit changes regularly during the implementation
sessions, but do not commit changes of work which they have been doing before
the session itself. Some groups commit rarely during the session and make a big
commit at the end of the session. We did some effort in classifying commit mes-
sages and noticed that students do often write messages which are either trivial
or even sheer nonsense. Further, we looked at how the usage of the system is
divided inside groups and found out that the activity of the most active user
in a group is significantly higher than what would be expected if each group
member would use the system equally much. As the last part of our analysis
we considered the types of files which students put under version control. We
concluded that if students are not told explicitly that they should not include
certain types of files, they will just do so.
232 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
Fig. 6. Fractions of the groups with a given percentage of commits performed by its
most active student
The Use of Distributed Version Control Systems ... 233
Fig. 7. Fractions of the groups with a given percentage of ’non-versionable’ files in
their repositories
With regard to the hypothesis put forward in the introduction, our findings
suggest that using a VCS as a submission management tool may result in stu-
dents adopting the tool as “just a required manner to return the assignments”
— instead of a professional tool by which collaborative and distributed work
is managed. Indications for this were that across all three courses studied the
version control system was not too evenly used by the team members and the
fact that the quality of the commit messages was quite low. While VCSs are
pronounced as useful course management tools in the literature, we would like
to note that professional use of VCSs requires support and demonstration of
their usefulness.
In our future undertakings, we could make a distinction between submissions
returns and use of VCS, and add VCS training to the beginnings of the courses.
Further, the existing classroom setup where there is sometimes only a single
computer for the whole group could be replaced with settings where each student
would use a separate computer. Performing these practical changes could reveal
whether a more intense and shared use of a VCS can be prompted among the
students. Promisingly, in our second-year course, training sessions were provided
and there were no observable commit peaks near the deadlines of assignment
phases but a rather constant commit curve. Further research could also point
out how effective the students can use the system and how the organization of
the group work influences the use of the system. It might be that some students
know very well how to use the system, but do not see any reason to use their
skills up to a full extent in the given settings.
Acknowledgments: We would like to thank the department of Mathemat-
ical Information Technology of the University of Jyväskylä for both financial
234 M. Cochez, V. Isomöttönen, V. Tirronen and J. Itkonen
and material support, without which this research would not have been possi-
ble. We would also like to thank the reviewers for their useful corrections and
suggestions, which greatly improved the quality of this paper.
References
1. Laadan, O., Nieh, J., Viennot, N.: Teaching operating systems using virtual appli-
ances and distributed version control. In: Proceedings of the 41st ACM technical
symposium on Computer science education. SIGCSE ’10, New York, NY, ACM
480–484 (2010)
2. Meneely, A., Williams, L.: On preparing students for distributed software devel-
opment with a synchronous, collaborative development platform. In: Proceedings
of the 40th ACM technical symposium on Computer science education. SIGCSE
’09, New York, NY, ACM 529–533 (2009)
3. Mierle, K.B., Roweis, S.T., Wilson, G.V.: CVS data extraction and analysis: A
case study. technical report utml tr 2004-002. Technical report (2004)
4. Reid, K.L., Wilson, G.V.: Learning by doing: Introducing version control as a way
to manage student assignments. In: Proceedings of the 36th SIGCSE technical
symposium on Computer science education. SIGCSE ’05, New York, NY, ACM
272–276 (2005)
5. Clifton, C., Kaczmarczyk, L.C., Mrozek, M.: Subverting the fundamentals se-
quence: Using version control to enhance course management. SIGCSE Bull. 39(1)
86–90 (March 2007)
6. Hartness, K.T.N.: Eclipse and CVS for group projects. J. Comput. Sci. Coll. 21(4)
217–222 (April 2006)
7. Liu, Y., Stroulia, E., Wong, K., German, D.: Using CVS historical information to
understand how students develop software. In: MRS 2004: International Workshop
on Mining Software Repositories. (2004)
8. Lee, B.G., Chang, K.H., Narayanan, N.H.: An integrated approach to version
control management in computer supported collaborative writing. In: Proceedings
of the 36th annual Southeast regional conference. ACM-SE 36, New York, NY,
ACM 34–43 (1998)
9. Glassy, L.: Using version control to observe student software development pro-
cesses. J. Comput. Sci. Coll. 21(3) 99–106 (February 2006)
10. Xu, Z.: Using git to manage capstone software projects. 159–164 (2012)
11. Milentijevic, I., Ciric, V., Vojinovic, O.: Version control in project-based learning.
Computers & Education 50(4) 1331–1338 (2008)
12. Rocco, D., Lloyd, W.: Distributed version control in the classroom. In: Proceedings
of the 42nd ACM technical symposium on Computer science education. SIGCSE
’11, New York, NY, ACM 637–642 (2011)
13. Tirronen, V., Isomöttönen, V.: Making teaching of programming learning-oriented
and learner-directed. In: Proceedings of the 11th Koli Calling International Con-
ference on Computing Education Research. Koli Calling ’11, New York, NY, ACM
60–65 (2011)
14. Tirronen, V., Isomöttönen, V.: On the design of effective learning materials for
supporting self-directed learning of programming. In: Proceedings of the 12th
Koli Calling International Conference on Computing Education Research. Koli
Calling ’12, New York, NY, ACM 74–82 (2012)
The Use of Distributed Version Control Systems ... 235
15. Isomöttönen, V., Tirronen, V.: Teaching programming by emphasizing self-
direction: How did students react to active role required of them? ACM Transac-
tions on Computing Education Research (accepted)
16. Isomöttönen, V., Tirronen, V., Cochez, M.: Issues with a course that emphasizes
self-direction, submitted. (2013)
17. Parzen, E.: On estimation of a probability density function and mode. The annals
of mathematical statistics 33(3) 1065–1076 (1962)
Comparative Analysis of Learning in Three-Subjective
Didactic Model
Aleksander Spivakovskiy1, Lyubov Petukhova 1, Evgeniya Spivakovska1,
Vera Kotkova1 and Hennadiy Kravtsov1
1
Kherson State University, 27, 40 RokivZhovtnya St., Kherson, 73000, Ukraine
{spivakovsky, petuhova, spivakovska, veras, kgm}@ksu.ks.ua
Abstract. The article theoretically shows transformation of modern didactic
model into three-subjective (Student - Teacher - Information and communica-
tion pedagogical environment). Active components of new subject, which are
the most evident in a learning process, are analyzed in the article. The require-
ment block of information and communication environment as a subject of the
educational process is described. The comparative characteristic of the main
components of traditional and innovative teaching systems is presented in the
article. The authors have made a comparative description of the main forms of
university studies in different didactic models: object-subject, subject-subject
and three-subject training. The measurement of cogency of each of these three
study subjects and their significance in the process of major educational opera-
tions (collection, processing, storage, transmission) in various forms of training:
lectures, practical classes and individual work were presented.
Keywords. Didactics, information society, information and communication
pedagogical environment, three-subjective didactics, forms of training organiza-
tion at university
Key terms. KnowledgeEvolution, KnowledgeManagementMethodology, Di-
dactics, KnowledgeManagementProcess, ICTInfrastructure
1 Introduction
Education is an institute of social experience transmission and human socialization in
society. Naturally it depends on the level of social development and labor market
needs.
Modern university education is in crisis, according to UNESCO specialists’ defini-
tion, it helpless and failure of modern education may bring to global problems of hu-
manity. These are irregular development of different countries in the context of glob-
alization, education inactivity caused by the relative conservatism of human resources
brings to constantly fast-moving knowledge renewal.
Weak sides of university education are the following students’ training instead of
Comparative Analysis of Learning in Three-Subjective Didactic Model 237
general cultural development, low professional motivation and responsibility, strict
regulation of students’ activities, provided graduates’ inactivity, not much attention to
the levels of training, etc. According to this fact, it’s said about global educational
crisis, the paradigm shift in pedagogical thinking [1].
We are going to trace the change of professional education at different stages of
society’s development to overcome the crisis of modern university education.
Down the ages human language existed primarily in form of sound speech. Its
main limitation was space-time limit: spoken word spread out the territory limited by
physical laws of sound and in form of material reality actually existed only while
pronouncing, straight after that passing into the history and vanishes in it. The era of
word was characterized by a certain lack of knowledge acquirement and the Institute
of transmission, as the main source of the word transmission process from one gen-
eration to another appeared to be a man.
The increase of information amount became the background for writing nascence
as it was difficult to keep the information in mind without losing its content. Writing
unlike the sound speech turned to be the technology of knowledge transfer.
The invention of writing (i.e. the possibility of fixing speech using a specially de-
veloped system of graphic symbols) allowed to transmit voice information to an
unlimited distance and extremely broadened its existence in time. Beyond dispute, the
appearance of writing created new additional conditions and opportunities to realize a
potential of human culture. But at the same time, writing leads to limitation and nar-
rowing of informational content of speech. An issue is that writing is a sign system
and it is shown as a representative of the signified so it reproduces only a half of
properties and meanings of what it means. In this case, word transfers only a part of
the properties and meanings contained in the "live" speech. Thus, written language is
actually completely lost in the so-called prosodic information contained in the "live"
speech that sounds. The case is that the graphical definition is losing information that
is expressed and transmitted in direct speech by means of phonetics which plays a
divisionary role.
Year 1450 AD (500 years ago) is marked by the appearance of new information
technology, the third one in a row. Only then printing technology appeared which we
consider a knowledge distribution technology. We call this phase an era of books.
Definitely the appearance of books allowed creation of an effective and mass educa-
tion system, to organize public libraries, to ensure the development of universities.
The appearance of books as a mean of transmitting knowledge, promoted the human-
kind’s achievement of those heights which it has now.
An important consequence of definite social development turned to be an under-
standing of purposeful activity of social skills transfer from one generation to another
as a connection between two organized activities – teaching and learning, their con-
crete reflection in the learning process. Humanity cumulative experience of the learn-
ing process has found expression in didactics, one of the pedagogy’s section that ex-
amines general theory of education and training. It is believed that the term was intro-
duced by German pedagogue Ratko in his lectures “Rahitiy’s summary of didactics
and art education”, meaning a scientific discipline which studies theory and practice
of teaching [2].
238 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
Efforts to make an educational process intelligently organized and purposeful are
presented in many Jan Komenskiy’s works, especially in "Great didactics," which
covers almost all the issues that present the subject of modern pedagogy. Jan Komen-
skiy was the first one who developed didactics as a system of scientific knowledge,
giving a reasoned exposition of principles and rules for children's education. He ex-
amined the most important questions of the learning theory: educational content,
teaching visualization principles, sequence of education, organization of class-and-
lesson system, etc.
Object-subject teaching relations between teachers and students of that time were
the most prolonged in pedagogy. The subject is the teacher who works actively to
educate students as his objects of influence through informative-educational environ-
ment (word, book, equipment). Schematically such didactic relationship is depicted in
Fig.1.
Teacher Student Teacher Student
Information and communi- Information and communica-
cation environment tion environment
Fig. 1. Schematic model of the object-subject Fig. 2. Schematic model of the subject-
relations subject relations
Gradual development of public experience of mankind has increased to such an ex-
tent that a person using only natural abilities was not capable to learn and operate with
informational resources. As the result, the person begins to use technological tools to
optimize working process with information. According to it, labor market no longer
corresponds to specialists’ "conveyor" training, as received mental vocabulary knowl-
edge quickly becomes obsolete, an employee must make a "decision" in unusual
situations. Naturally, a student becomes an equal subject of the educational process.
The transformation of an object into a subject in the educational process is the result
of the democratization of education, dissemination of differentiation and individuali-
zation of education. Schematically, the new relationship is depicted in Fig. 2.
Intensification of information processes, introduced into science, economics, pro-
duction, requires the development of new models of education, a variety of informa-
tion and communication environments in which people could reveal their creativity
fully, develop skills and cultivate a necessity for self-improvement and responsibility
for their education and development.
The traditional paradigm considered education as training of younger generation to
work and life by consuming material valuables created in other areas. The new para-
digm foresees independent values in education.
Comparative Analysis of Learning in Three-Subjective Didactic Model 239
2 Innovative methodical system
The purpose of creating a new education paradigm is to provide conditions for educa-
tion, training and development for independent, smart person to satisfy the require-
ments on a market economy, capable to improve continuously his own level of
knowledge and culture, integrated into the global informative space.
Thus, today, we are talking about innovative methodical system which unlike tradi-
tional one, corresponds to professional education demands in an informative society.
The comparison of the main components of these two systems is presented in Table 1
[3].
Table 1. Comparative characterization of traditional and innovative teaching systems.
Name of
Traditional methodical system Innovative methodological system
component
Learning Seizure and adoption of educa- Provide students with knowledge,
objectives tional material. skills and practice.
Provide students with knowledge, Creation of modern information and
skills and practice. communication teaching environ-
ment.
Purposeful development of creative
self-sufficing person.
Formation of professional competence,
leadership skills, ability to work in
group.
Principles of The scientific character principle. The principle of the activity learning
learning The principle of systematicity and environment.
consistency. The principle of organic unity be-
The principle of visibility. tween the changing requirements at
The principle of studying direction labor market and conserved features
in accordance with issues of edu- of the educational system.
cation, training and development. The principle of necessity for contin-
ual self-study.
Contents of Classical learning, techno- An integrated approach to fundamen-
training cratic one. tal and applied activity aspects of a
specialist-to-be.
Study Reproductive, explanatory, illus- Problem-search, research.
methods trative.
Study Visual tools. The teacher’s word - Facilities. Information and communi-
means for knowledge transfer, books, cation technologies. Hypertext, mul-
movies, tape, training devices, timedia training materials. Databases
pictures, maps, tables, machines, for educational purposes. Networking
devices, models, collections, tools, means for videoconferencing and
and historical schemes, charts, video lecture. An effective system of
240 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
Name of
Traditional methodical system Innovative methodological system
component
diagrams, etc. Technology. Video- monitoring training activities. Re-
recordings, radio and television, mote devices for self-work. Com-
filmstrips, slides, transparencies, puter testing in on-and off-line
projectors, televisions. modes.
Study Lectures, seminars and Dispute, seminars, conferences,
forms practical lessons. "round table", symposium, debates,
colloquium, distance learning, teach-
ing and business games, role-play
game.
Control External process operations con- Strict current control of individual
forms trol within strictly defined rules is learning of each student by means of
dominated. testing in on-and off-line mode.
A teacher assessment result (flow, Rating control knowledge.
final control) is dominated. Creating an effective environment
Lack of balance between control according to Jean Piaget for easy
and self-control. convenient self-organization that
Lack of effective control for indi- motivates students in learning activi-
vidual learning methods of each ties.
student.
In addition, society today has faced the phenomena which require answers:
1. Teacher has lost the monopoly on knowledge;
2. Students have unlimited access to information resources;
3. The phenomenon of "red shift" in expanding informative and communicative
space;
4. Availability of qualitatively and quantitatively different ICT competencies of
young and older generations.
For that matter, an educational paradigm transforms, which is characterized by the
following principles:
Globalization of knowledge, free access to educational resources
Integration of learning resources
Organization of global educational audiences
WEB-multimedia presentation of educational resources
Multilingual educational space
Asynchrony of modern models for learning management
Harmonization of social and educational environment
Formation of social identity of information system
Divergence in the implementation of their own educational way
Thus, an evolution of modern education, information studies, mass computeriza-
tion of educational establishments, constant upgrade of hardware, and development of
Comparative Analysis of Learning in Three-Subjective Didactic Model 241
computer networks, expanding of personal computerization of society, increasing of
software products designed for use in an educational process – these are conditions
that create new information and communication pedagogical environment (ICPE).
This environment constantly and aggressively increases student’s motivation to con-
sume content that circulates in it, creating a new didactic model – three-subjective
relations, which include three subjects of study - students, teachers and an environ-
ment.
However, is it legitimate to consider ICPE a possessing equal rights subject for
learning along with a teacher and a student?
3 Model of three-subjective relations
Consideration of information and communication teaching environment as a subject,
in our opinion, is possible because its components are not only technology but human
resources as well, which continuously update them at the constantly growing speed.
In this sense, it is necessary to point out an existing qualitatively new learning envi-
ronment as opposed to which one that was 15-20 years ago. The question deals with
the obtaining of today's educational environment the status of an equal partner. Sir
Ken Robinson in The Third Teacher (2010) says, “The physical environment of the
building is critically important in terms of curriculum” [4].
Within this approach, we implement an important target triangle: a natural integra-
tion of teaching, research and labor market needs. After all, ignoring the environment
as a subject of education, we will prepare specialists for inadequate reality.
The inevitability of the transition of the education system to consider three-
subjective relationship is reflected in the following three stages of didactic changes:
Stage I – the subject-object instruction (a teacher provides students with knowl-
edge). Characterized by one-dimensional linear model, the volume of processed data
– megabytes;
Stage II – subject-subject didactics (a teacher and a student are equal competent
training partners). Characterized by two-dimensional polylinear model, the volume of
processed data-gigabytes;
Stage III – three-subjective pedagogy (a teacher – a student – ICPE). The interac-
tion of all subjects of the learning process (a teacher – a student – ICPE) obeys to the
common goal which is formation of a competitive specialist and is characterized by a
three-dimensional nonlinear model, the volume of processed data – terabytes.
Thus, we have the right to talk about three-subjective didactics as one of the areas
of pedagogical science of the most general regularities, principles and means for or-
ganization of studying, providing a firm and conscious assimilation of knowledge and
skills within peer relations pupil (student), teacher (teacher) and information and
communication teaching environment.
It is important to underline that in this process, status and general condition of
those who learn and teach and ICPE are constantly changing. In this context, we un-
derstand these learning activities with assimilation of knowledge and skills, and
teaching - the knowledge message or source of knowledge to students, as well as in-
242 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
struction on ways and methods of work, coordinating training activities, particularly
organization of active forms (discussion, round table, project activities, etc.) and
monitoring of students mastering knowledge, skills and experience obtaining. Unlike
traditional views, we consider, that it’s necessary to introduce the one who teaches
into the learning process, the changes that are ICPE (for example, by means of pub-
lishing of educational materials in the Internet). We have to mention, that new innova-
tive forms of teaching activity are connected remotely or, as they say, distance man-
agement software training activities, both in time and space.
Within this definition naturally occurring three-subject relations, which we under-
stand as the continuous and constant (both in space and time) interactions between
students, teachers and information and communication pedagogical environment di-
rected for satisfaction of students educational needs (Fig. 3).
Student Teacher
• has some educational needs • forms relevant competence
•requires to reveal and develop • coordinates students’ self-
personal abilities development
•strikes for information culture • coordinates students’ thorough
development
Information and communication pedagogi-
cal environment
Fig. 3. Schematic model of three-subjective relations
As we are examining an environment as a separate element, it’s important to men-
tion its operation characteristics, which are the most evident in the learning process:
environment constantly and more aggressively increases motivation of the younger
generation for content use that circulates in it
environment provides access to resources at any convenient time
environment has comfortable, flexible, friendly, intelligent service, that helps peo-
ple to find informational resources, data or knowledge they need
environment is not a negative emotional one, it corresponds people’s demands as
much as it is needed
environment permanently is filled up with information, data, knowledge with a
constantly increasing speed
environment offers an opportunity to organize practically free, time convenient
contacts between any number of people to provide suitable and flexible informa-
tion exchange (in any form) between them
Comparative Analysis of Learning in Three-Subjective Didactic Model 243
environment, step by step, standardize, and then integrates the functionality of all
previous, so-called traditional ways of receiving, storing, processing and presenting
of the required information, data and knowledge to mankind
environment undertakes more and more routine operations connected with humans
operating activities (which is one of the greatest challenges for humans to expect in
the future - "the more commissions - the more responsibility - the greater risks re-
main without resources")
environment receives more and more control over the data, and operational man-
kind’s activities [3]
Due to our three-subjective didactics, we can answer the above-listed vital ques-
tions connected to modern educational system:
the teacher’s role and place in the new didactic model
correlation between virtual and visual forms of subjects’ relationships in didactic
system
development of technological management providing rights for subjects of didactic
system to login informational resources
organization of modern control systems over learning activities
assurance of the organic unity between changing requirements of labor market and
conservative educational system potential
organization of modern, and most importantly, systematic and constantly active
system of re-training and professional qualification upgrade of teachers
Step by step, it is getting clear that technology that produces modern industry, to-
day, not only affects the technology transfer of knowledge, but in fact, it determines
qualitatively new forms of its organization for their mastering. At this stage, we can
see the following problems:
1. Heterogeneity of distribution of computer and communication facilities
2. Huge differences in the process of training and constant re-training of staff, both
academic and administrative ones
3. Inertia of education system
4. Constantly growing volume of technological renewal of learning environment that
includes all the tools, both for teacher and learners
5. Imposing of different learning paradigms that make substantial confusion in the
teachers’ presentation of their new role in the process of knowledge transmission,
development of abilities and skills
6. Stereotype of the philistine attitude to pedagogy in whole, as a descriptive section
of human knowledge, in which every citizen is a knowledgeable expert
7. Absence of formal systems which describe different models of learning [3]
Active learning environment contains the following units, which are procedural,
substantive and control. The environment begins to play a more important role and
assumes some part of teacher’s functions. There is no doubt that, certain requirements
must be done in the process of setting up a proper learning environment, which will
provide active learning environment. Working with the program, both a student and a
teacher will be limited by a system of actions, which was laid out in the program,
244 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
that’s why development of the system requirements of ICPE is very important. Ac-
cording to our research, information and communication learning environment can
serve as the subject of the educational process if it meets the following group re-
quirements:
1. Hardware requirements: multimedia computers in classrooms are networked with
the obligatory access to the Internet resources. In addition, an important aspect
creates opportunities to access educational electronic resources (Wi-Fi technol-
ogy) for students in any convenient place, for example, library, dormitory, canteen
etc.
2. Software requirements: software environment should resolve security issues (reg-
istration, personalization, delineation of access rights to get to resources), to be in-
tegrated (all educational components should be in its natural form), easy for ex-
ploitation, filling and modification, to provide opportunities for interaction, com-
munication, monitoring for learning process, to contain an output mode out of the
complicated situations (expert), to offer opportunities for distance learning (on-
and off-line modes).
3. Academic requirements refer to methods of filling information and communica-
tion teaching environment.
4. Social demands. Special attention should be paid to a specified group of claims
which, in our opinion, contains cultural, ethical and legal aspects, because users of
information and communication pedagogical environment create some commu-
nity. First of all, it is about the rules of communication in the network and use of
the reworks of other authors.
Requirements to Human Resources. Construction of the educational process on
the basis of information and communication technologies implies specialists- pro-
grammers and accordingly well-trained teachers.
ICPE correspondence to these requirements can be achieved by using management
system of the quality of educational information resources [5].
Introduction of new subject of learning process naturally transforms existing ele-
ments of training, including forms of teaching at higher educational establishments.
Today, educational resources are open and distance forms for studying are actively
developing and integrating into traditional forms of teaching: lectures, workshops,
laboratory classes, independent, individual work of students, forms of control. Let's
try to analyze basic traditional forms of training organization in different didactic
models (Table 2).
Table 2. Forms of learning in didactic models
Subject-object study Subject-subject study Three-subject study
Teacher
The source of educational A teacher presents difficult A teacher and students in the
information is a teacher; educational material, stu- debate form discuss problematic
students are forced to put dents selectively put down issues due to free access to open
down a limited amount of the information that is nec- lecture and other information
Comparative Analysis of Learning in Three-Subjective Didactic Model 245
Subject-object study Subject-subject study Three-subject study
information, static visibil- essary for each personally, sources. Students write down
ity is used additionally. use additional sources, in- the required information at will.
cluding the Internet. Dy-
namic visibility is domina-
tive.
Practice
Reproductive methods of Part-search training methods Search and creative methods are
teaching material devel- prevail. directed at forming experience
opment are used. of training materials, particu-
larly under unusual circum-
stances.
Independent work
It consists of lectures, Studying unwrought amount The main part of teaching mate-
practical exercises execu- of teaching material. rial is studied individually.
tion.
Forms of control
A control requires pres- Students’ readiness to use Monitoring can be conducted
ence of a teacher who received knowledge in con- without teacher’s presence, and
relates a student's knowl- dition of life situation is also the result – unconventional
edge with the volume of under control. approach and creative thinking
lectures material. of students are estimated.
4 Measurement of the importance of training subjects
However, is ICPE a significant, important subject of learning in practice of University
operation? The theoretical conjectures study was conducted at the Faculty of pre-
school and primary education of Kherson State University in order to confirm or re-
fute it. The research required a questionnaire of future primary education teachers, as
they acquire an integrated system of philology, humanities, exact, natural and artistic
sciences, which, in our opinion, reduce a risk of results’ obtaining only from certain
cycle of training. The main task of the questionnaire is to evaluate the significance of
subjects of modern educational process, including ICPE.
Determining the validity of each of the three subjects of the educational process
was made by expert evaluation method. 27 qualified experts (university teachers,
graduate students, methodologists) joined the independent expert committee.
To define a point of evaluation for each subject Delphi method (for members of the
expert committee conditions for an independent individual work were created) was
used [6]. Maximum and minimum estimates depended on a number of subjects, in
which, there are three. Thus, minimum score for one of three components –1 point, an
average score – 2 points and maximum – 3 points. Then, the statistical processing of
the results, which were presented to experts for final approval, had been conducted.
The cycle of expertise was repeated three times.
246 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
Below are the results of an independent expert committee (Table 3).
Table 3. Determination of cogency of training subjects (V)
Subjects of the educa- Number of points
∑ V
tional process 1 2 3
Teacher 12 8 7 49 0,30
Student 2 10 15 67 0,41
ICPE 12 11 4 46 0,29
According to the results of expert reviews cogency V (in fractions of a unit) for
each of these three specified subjects, according to experts, is approximately the
same, with a slight advantage "student" (0.11 larger compared to "ICPE" and 0.12
larger compared to the "teacher").
Results summarizing the data are shown in Fig. 4.
0.29 0.3
Teacher
Student
ICPE
0.41
Fig. 4. The importance of the subjects of the educational process
214 students, as the most significant subject of didactic system according to ex-
perts’ definition, were asked to rate on a 10-point scale the importance of the three
subjects of the educational process: students, a teacher and ICPE – in the process of
operating with information (collecting, processing, storing, transmission) in various
forms of training organization: lectures, practical classes and independent work. To
do this, students were asked to determine the importance of each component of didac-
tic models: Student - Teacher - ICPE a five-point scale (1, ..., 5).
Results of the student’s questionnaire are shown in Table 4.
Assessment Eij (i, j = 1,2,3) for each i-component of the didactic system operations
in terms of j-forms of training organization is given by (1):
Eij=Kij1+Kij2+Kij3+Kij4, (1)
where Eij– total score in terms of transactions weight, Kijk - i-score weighting compo-
nents didactic system, j-teaching forms and k-rate transactions, %.
The overall assessment of Vi (i = 1,2,3) for each component of the didactic system
is given by (2):
Comparative Analysis of Learning in Three-Subjective Didactic Model 247
Table 4. The results of the student’s questionnaire
Components Student Teacher ICPE
Form of Indicators
training
organization points % points % points %
Collecting 1182 5,5 2187 10,1 334 1,5
Processing 2151 10,0 2630 12,2 2411 11,2
Lecture
Storing 2625 12,2 1455 6,7 2402 11,1
Transmission 830 3,9 2004 9,3 1355 6,3
∑ 6788 8276 6502
E 31,6 38,3 30,1
Collecting 1674 7,7 2006 9,2 1235 5,7
Processing 1885 8,7 3121 14,3 815 3,8
Practice
Storing 1241 5,7 1663 7,6 2421 11,1
Transmission 2178 10,0 1198 5,5 2322 10,7
∑ 6978 7988 6793
E 32,1 36,6 31,3
Collecting 2214 9,7 2119 9,3 2033 8,9
Processing 2366 10,4 2882 12,6 1938 8,5
Independent
Storing 2154 9,4 2007 8,8 2013 8,8
Transmission 994 4,4 384 1,7 1703 7,5
∑ 7728 7392 7687
work
E 33,9 32,4 33,7
Vi= (Ei1+Ei2+Ei3)/3. (2)
Let’s analyze the results.
It is generally known that, lecture – is the main form of teaching, prepared for the
adoption of theoretical material. Table 4 shows that while gathering information dur-
ing lectures (17.1%), the most significant entity of the educational process is a teacher
(10.1%), 5.5% of operation is performed by a student, 1.5% – ICPE. It’s explained by
identification of the content and material of lectures, in its selection, the main role is
occupied by a teacher, but the lecture provides not passive acceptance of students’
knowledge but their active involvement into the learning process, preparation for
lectures, which is provided with ICPE use. We should note that active cognitive activ-
ity of students during lectures is possible for basic training, which includes familiari-
zation with the theme of the lecture and its plan, the main content of the theme for the
tutorial, content repetition of the previous themes etc.
According to the survey results, information processing on the lecture (33.4%) sub-
jects’ contribution is approximately the same: 12.2% - Teacher, 11.2% - ICPE, 10% -
Student. This is because the teacher coordinates educational information processing,
and an active entity involved in this process may be a student. ICPE activity due to a
shift in emphasis onto the use of methods and means of processing students - from
note-taking information material: full or theses synopsis for computer processing of
the information received.
248 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
Storing educational information of the lecture (30%) between the subjects of the
educational process was distributed: 6.7% - Teacher, 11.1% - ICPE, 12.2% - Student.
According to received questioning results, transfer of educational information of
the lecture (19.5%) is implemented by a teacher (9.3%) and ICPE (6.3%), although
students (3.9%) provide additional information, interesting facts and problematic
issues. The task of the teacher, at this stage, is to transfer the adapted information
disclosing a nature of scientific concepts, genesis of scientific theories, ideas, etc.;
aggregated information is transmitted using pedagogical software, e-presentations,
etc.
The results of the distribution of three-subjects training, in the process of operation
with information during a lecture are presented in Fig. 5.
Fig. 5. The significance of the subjects of the educational process during operations with in-
formation on the lecture
Thus, the most important subject of the lecture organization according to students
is a teacher (38%), a student (31.6%) and ICPE (29%) provide processing and preser-
vation of educational information. In general, during the lecture principal place of
work with information take processing operations (33,44%) and storing (30%), fol-
lowed by transfer (19.5%) and collection (17.1%).
Let’s analyze the data from Table 4 according to the importance of the subjects of
study during the preparation and conduction of practice. As it is known, practical
lesson is a class that involves organizing teacher’s detailed study of individual theo-
retical positions discipline and development of skills in their practical application by
individual performance to related tasks.
Analyzing the data in Table 4 concerning the collection of information (22.6%)
during the practical sessions was revealed that students’ contribution is 7.7%, but
teacher’s and ICPE respectively 9.2% and 5.7%. Comparing with a lecture, students’
activity increased by 2.2%, due to test theoretical knowledge of students, develop-
ment of skills based on acquired knowledge and, as a result, a detailed collection of
information for further processing.
In the process of collection (22.6%) and processing (26.8%) of the information
during preparation and practice, according to students, a teacher and a student have
the greatest significance, which is confirmed by received data: respectively (9.2% and
14.3%) , (7.7% and 8.7%). This is due to students interest in learning, deepening and
refinement of knowledge, developing skills, primary accumulation of experience,
Comparative Analysis of Learning in Three-Subjective Didactic Model 249
professional motivation and, consequently, activity in learning. Significance of ICPE
is gradually increasing, as it is evidenced by statistics data, in storing and information
transmission: 11.1% and 10.7%.
Visually, the results are presented in Fig. 6.
Fig. 6. The significance of the subjects of the educational process during operations with in-
formation during the practice
Thus, a teacher is the most important subject of practical training organization
(36.6%), although a student (32.1%) and ICPE (31.3%) are equal subjects. Generally
during practical classes the importance of operations with information is as follows:
processing (26.8%), transmission (26.2%), storing (24.4%) and collecting (22.6%).
Let’s analyze the importance of training subjects in the process of student’s indi-
vidual work organization. As you know, independent work of a student is a primary
mean to master academic material at a time, free from mandatory training sessions.
Leading role in collection, processing, storing and transmission of material belongs
to a student (33.9), according to the relevant data: 9.7% 10.4% 9.4% 4.4%. This is
primarily due to the students' understanding of the importance of having theoretical
knowledge, development of skills, and accumulation of their own professional experi-
ence and, as a result, operations with the information according to the educational
goals. Practice has proved that the most active in independent work will be a student
who is more motivated to master for his future profession. The result of questioning is
the importance of teachers is on average 32.4%, ICPE - 33.7%, and an independent
educational-cognitive students' work is task-teacher, under his leadership, but without
his direct involvement but widespread use of information and communication teach-
ing environment. The importance of business education in independent work is shown
in Figure 7.
So, according to students’ definition, important subjects of independent work are
all three components of the didactic system - Student (33.9%) and Teacher(32.4%),
and ICPE (33.7%). During the independent work with information, transaction proc-
essing occupies a principal place (31.5%), then collection (27.9%), followed by stor-
ing (27%) and afterwards transmission (13.6%).
250 A. Spivakovskiy, L. Petukhova, E. Spivakovska, V. Kotkova and H. Kravtsov
Fig. 7. The significance of the subjects of the educational process during information opera-
tions in class work
The analysis of the results of the survey showed that according to students' all three
constituents are important and significant components of the didactic Student-
Teacher-ICPE system. This is the statistics of indicators weight components: Student
(V1 = 32,5%), Teacher(V2 = 35,8%), ICPE (V3 = 31,7%). It’s important to underline
that received students’ survey data correlate well with similar data of experts’ assess-
ment of component’s importance of the didactic system. The student is a significant
subject of a teaching process at the University as learning outcomes largely depend on
its intended acquisition of trade. Proof that serve high levels of significance to stu-
dents in maintaining and processing information during lectures, collecting, process-
ing and transmitting information during practice, collecting, processing, storing in-
formation during independent work. Major indicators of the importance of the teacher
can be seen during collection, processing and transmission of information during the
lecture, which is determined by specifics of this type of training sessions - teaching
theoretical material. During practical sessions and independent students’ work the
teacher has the greatest indicators of the importance of information processing.
ICPE’s significance is high during operations with information and practical lessons
in the process of information preserving, as for in-class, ICPE acts as an equal-right
subject of the educational process. This is because ICPE provides access to informa-
tional resources at any convenient time, quickly and easily enables to find all neces-
sary information, provides flexible and convenient information sharing between stu-
dents. However, according to average data ICPE significance inferior teachers’ im-
portance, as a teacher manages the studying-cognitive students’ activity, coordinates
their independent improvement of knowledge, skills and abilities. It should be noted,
that due to ICPE systematical involving in learning process the role of it as new sub-
ject will improve gradually because of improving learning outcomes.
Summary results of the survey are presented in Fig. 8.
Comparative Analysis of Learning in Three-Subjective Didactic Model 251
Fig. 8. The importance of the subjects of the educational process during operations with infor-
mation
5 Conclusions
Thus, the analysis of the scientific literature, theoretical and experimental study on
transforming learning into different didactic models showed that information and
communication pedagogical environment is an important subject in the process of
learning at the University; ICPE transforms traditional subject-subject model of study
into three-subject one, directly affecting and slightly changing the role and function of
other subjects of study, partly fingering their functions itself, particularly in transient
conditions while performing operations with information on various forms of training.
References
1. Fokin, J. G.: Competencies of Education in University. Academy, Moscow (2002)
2. Kendall, H. L., Sugimoto, R. A.: The Didactic Theory of Wolfgang Ratke. California State
University (1976)
3. Petukhova, L. E.: Theoretical Bases for Training Primary School Teachers in Information
and Communication Teaching Environment. Scientific Monograph. Ayilant, Kherson
(2007)
4. The Third Teacher: 79 Ways You Can Use Design to Transform Teaching & Learning by
OWP/P Architects, VS Furniture, Bruce Mau Design, Abrams (2010)
5. Kravtsov, H. M.: Design and Implementation of a Quality Management System for Elec-
tronic Training Information Resources. In: Ermolayev, V. et al. (Eds.) Proc. 7-th Int. Conf.
ICTERI 2011, Kherson, Ukraine, May 4-7, 2011, CEUR-WS.org/Vol-716, ISSN 1613-
0073, pp. 88-98, CEUR-WS.org/Vol-716/ICTERI 2011-CEUR-WS-paper-6-p-88-98.pdf.
(2011)
6. Rowe, G., Wright, G.: Expert Opinions in Forecasting: the Role of the Delphi Technique.
In: J.S. Armstrong (Ed.) Principles of Forecasting – a Handbook for Researchers and Prac-
titioners, pp. 125-144. Kluwer Academic Publishers, Boston, MA (2001)
Conception of Programs Factory for Representing
and E-Learning Disciplines of Software Engineering
Ekaterina Lavrischeva1., Artem Dzyubenko1. and Andrey Aronov1.
1
Taras Shevchenko Kiev National University, Kiev, Ukraine
lavryscheva@gmail.com, asmer@asmer.com.ua
Abstract. The paper presents a new idea of knowledge representation for stu-
dents studying software engineering by developing artifacts and software com-
ponents accumulated in libraries or repositories for further reuse. The idea is
based on the concept of assembly line by V. Glushkov, further elaborated by
Soviet and foreign specialists (A. Ershov, V. Lipaev, J. Greenfield, G. Lenz, Y.
Bai, M. Fowler). The paper introduces the elements of program factory: reus-
able components, their interfaces, and assembly lines for designing and assem-
bling complex software products from components. It is shown that modern op-
erating environments provide prerequisites for such factories. The students’
program factory, implemented at Taras Shevchenko Kiev National University,
is then described. The technologies behind the factory, as well as its goals are
studied.
Keywords. Artifact, reusable component, applied system, interface, assembling
line
Key terms. Methodology, Technology, Process, Qualification
1 Introduction
The paper presents the elements of programs factories: reusable components (RC) and
their interfaces, assembly lines for designing and assembling complex programs from
RCs. It is shown that modern operating systems provide tools for creating specialized
programs factories (MS.NET AppFabric, SOAFab, SCAFab, IBM VSphere, CORBA,
etc.). Factories are built by different commercial structures for software product de-
velopment, as well as for studying purposes (e.g., the programs factory implemented
in Taras Shevchenko Kiev National University for the assembling artifacts from dis-
ciplines that are studied at the university). The purpose of such factories is to study
computer science, software engineering, programming technology and software sys-
tems with electronic textbooks, develop applied projects and thus train highly compe-
tent professionals in software industry.
Over last decades a huge amount of various programs has been accumulated in the
informational world that may be used as end products for complex programs devel-
Conception of Programs Factory for Representing and E-Learning Disciplines … 253
opment. Therefore, a new approach has been formed in programming, namely reus-
ability – reuse of ready-made software resources (reuses, assets, services, compo-
nents, etc.), hereafter referred to as reusable components (RC). This term is used in
informational world to represent new knowledge acquired over researches in certain
fields of computer science. Being needed for somebody, it may be used in solving
certain problems concerning similar artifacts as well as for development of new soft-
ware systems (SS), applied systems (AS) or software product families (SPF).
All software artifacts and RCs may be stored in public warehouses (libraries, re-
positories) that may be searched by professionals to identify the necessary data for
implementation in their own research activities. That is, reuse of ready-made re-
sources becomes a capital-intensive activity in the field of software engineering and it
is particularly important that universities possess so-called factories for scientific
artifacts, programs and RCs needed by other students and professionals. With these
factories, students may participate in industry development of scientific artifacts for
mass use [1–3].
Based on such innovative ideas, Prof. E. Lavrischeva proposed fourth-year stu-
dents at KNU to establish the first programs factory over the course of theoretical and
practical labs on software engineering. This factory is focused on artifacts, software
development, and repository maintenance. Students’ programs factory operates on the
web site (http://programsfactoty.univ.kiev.ua) since December 2011. The web site has
been visited by more than 5,000 people – students, scientists and teachers.
2 Establishing Programs Factory
History of Software Industry in USSR. An idea of industry for computers and sup-
porting software has been formulated by Academician V. Glushkov at Cybernetics
Institute of NAS of Ukraine in 1960s-1970s. Under his guidance, a family of small
computers ‘Mir’ (1967–1975) has been developed together with a language of ana-
lytical and formula transformations for solving differentiation, integration and for-
mula calculus problems. In addition, other computers (Dnepr, Dnepr-2, Kyiv-67, 70,
macro-conveyor etc.) have been elaborated with auto code-typed programming lan-
guages (PL) to develop information processing programs and automated management
systems (AMS) for various organizations and enterprises. For their development, the
problems of software quality and increase of production of reusable components have
been investigated with computers at state institutions [4–6].
In 1975 V. Glushkov first formulated the concept of assembling conveyor consist-
ing from technological lines for software products (SP). The core of Glushkov’s para-
digm is to accelerate the transition from programming as an art to industrial methods
of SP production in order to solve various economical, business, scientific problems
with automated management systems.
Then (1978), software development as joint scientific-technical production and the
projects requesting for automating SP creation have been decreed. The first pilot fac-
tory for software engineering was established for mass production of various AMSs
(1978, Kalinin); but, because of lack of ready-made programs and immaturity of pro-
254 E. Lavrischeva, A. Dzyubenko and A. Aronov
gramming technology for industry production, the factory had lasted for two years
and was closed [7]. Nevertheless the experiments aiming to elaborate programming
automation tools were lasting until the collapse of the USSR.
V. Glushkov’s idea concerning programs factories is now running at the several
industrial factories explored by different authors theoretically and in practice [8–13]:
Conveyor by K. Czarnecki and U. Eisenecker
Software factories for assembling applications by J. Greenfield, K. Short et al.
Continuous integration by Martin Fowler
EPAM assembly line for building various types of software, improving software
quality and reducing risk
Automated assembling of the multi-language programs in heterogeneous environ-
ments by Y. Bai (VC++, VBasic, Matlab, Java, Visual Works, Smalltalk and oth-
ers)
Command development and assembling programs for software projects in MS
Visual Studio Team Suite based on contracts
G. Lenz’s program factory utilizing UML in .NET
Assembling, configuration and certification of global scientific-technical software
on the European Grid factory
Compositional (assembling) programming by E. Lavrischeva for developing soft-
ware products from reuses, services, artifacts, and so on
Experimental KNU factory
Careful analysis allowed singling out the key components of programs factories
[13]:
Prepared software resources (artifacts, programs, systems, reuses, assets, compo-
nents, etc.)
Interface as a mediator between two components, containing passport information
for heterogeneous resources in a certain specification language (IDL, API, SIDL,
WSDL, RAS, etc.)
Operating environment with system facilities and tools supporting assembly lines
with heterogeneous software resources
Technological lines or product lines for mass production and assembling products.
Methods of product development
The elements listed are the fundamentals of SP production industry at the factories
that operate at major foreign software companies such as Microsoft, IBM, Intel, Ap-
ple, Oberon etc. At the KNU students’ programs factory, these fundamentals are im-
plemented as certain lines for programs and scientific artifacts, which are the best
among student-developed programs factories.
The authors of the KNU programs factory consider it an integrated infrastructure
for organizing production of mass usage SPs that are needed for customers and users
from the fields of computer science, state government, commerce etc. The factory is
equipped with technology lines (TL) or product lines [7, 12], as well as a collection of
products, tools and services needed for automated processes execution over these
Conception of Programs Factory for Representing and E-Learning Disciplines … 255
lines in modern operational environments (MS.NET, IBM, Sun Microsystems and so
on).
Web Site of KNU Programs Factory. The main objectives for the web site are:
improving students’ skills in software development using the system for exchange of
KNU students’ certified software products and scientific artifacts; increasing SP qual-
ity and reliability; and learning to support the methods of SS industrial production.
The web site presents lifecycle models, SP building lines, the line for program pro-
duction with the help of MS.NET platform, and examples of student programs.
Obligatory requirements are maintained during certification to store software products
in the repository of the factory (the library pool).
The main activities on the factory site are:
Organizing program and artifact development
Familiarizing students with tools and methods for program and SS development
Representing students’ software products in the repository
Citing excerpts from articles and textbook materials concerning various disciplines
The factory is equipped with tools that allow broadening its functionality by specify-
ing new software artifacts and storing them in the repository for further reuse. Each
artifact is uniformly documented based on WSDL, IDL standard used in Grid global
project.
3 Factory Lines and Components
Development of Technological Lines. Technological lines are created at the techno-
logical pre-production stage [7, 12] before SS production. They include activities for
designing the TL scheme from processes and actions that determine the processing
order of SS elements with appropriate technological modules (TM) or programming
systems. The basic requirement of TL engineering imposed on production of pro-
grams and components is to assemble TLs from lifecycle processes meeting problem
domain goals using standard tools, TMs, and the system of regulatory documents. The
TL is then supplied with ready-made components, tools and instruments that generate
and implement specific functions or elements, as well as the management plan for
processes for changing states of the elements and providing quality evaluation [13-
16].
The RC model for component-based development has the following specification:
RC = {T. I, F, Imp, S}, (12)
where T is type, I is interface, F is functionality, Im is implementation, and S is inter-
operability service.
Basic operations over components are:
Specifying components and their interfaces (pre- and post-conditions, which must
be satisfied by the caller) in such languages as IDL, API, WSDL, etc.
256 E. Lavrischeva, A. Dzyubenko and A. Aronov
Maintenance of components, reuses and artifacts in the component repository for
search, change and future integration into applied systems
Integration of components into applications, domains, applied systems, software
product families, etc.
All aspects of RC development and their usage in SP and software product families
are goals for reusability disciplines and building material for these systems.
Applied system is a collection of software means (or functions of SS), including
general tools (DBMS, protection systems, system services, etc.), constructed subsys-
tems or components together with the tools for marshalling from one AS to another
[14].
Product Lines at SEI. Product lines and product family (PF) is defined in
ISO/IEC FDIS 24765:2009 (E) – Systems and Software Engineering Vocabulary:
“Product line is a set of products or services that share a common, managed set of
features satisfying the specific needs of a particular market segment or mission and
that are developed from a common set of core assets in a prescribed way. Synonym:
product family”.
SEI professionals propose the two models for representation of activities for SS
development, namely engineering and process ones.
The engineering model assumes the three activities, namely RC development, PF
development through RC configuration and management of the both above activities.
The activity for RC development presupposes PF scoping and production planning
of SS collection accounting for the context of SS usage, production limitations and
the chosen strategy. The PF development activity includes designing each specific SS
implementation based on the set of developed RC, and building software systems
according to the PF implementation plan. The management activity is based on bal-
ancing activities for RC development and PF maintenance tasks, and includes both
organizational and technical management.
According to the process model, a set of processes is performed at the two levels,
namely domain engineering, being also referred to as the development “for reuse”,
and application (or SS) engineering – the development “with reuse” [15]. The last one
is performed over assembly line using ready-made RCs to shorten time and increase
SS availability. Therefore, configuring the product family from RCs according to the
specific requirements and needs of a particular market segment is the final one in the
cycle of production activities.
4 KNU Students’ Programs Factory
From the theoretical standpoint, program factories are based on the assembling con-
veyor that includes a collection of various more or less complex production lines for
software artifacts, programs and RCs. Conveyor lines contain process execution using
system tools or technological modules that automate process execution for obtaining
interim or final results.
From the perspective of information technology, the factory provides the data
processing toolset for the transition from individual programming of particular re-
Conception of Programs Factory for Representing and E-Learning Disciplines … 257
sources to the industry of mass-usage SP. The factory increases SP development pro-
ductivity during each lifecycle process due to use of RCs that possess the necessary
functionality with due quality guaranteed by their developers. The assembling (com-
position, configuration) line may benefit through reduction of efforts because of use
of readymade artifacts or RCs stored in the repository.
Home Publications Learning Repository Contacts Link
Searching for RC Select Platform Searching for manuals Select Manual (if necessary)
All Platforms All Books
Select OS Searching for publications Select category (if necessary)
All Platforms All category
Input text for Searching
Search
Recent Reusable Components (RC)
Repository
Learning
Shingles algorithm Home Literature
Dzubenko Artem
Moxiecode System “Software Engineering” Manual
Author: Lavryscheva Katerina Myhailovna
Aronov Andriy
Dzubenko Artem
Generic Line for particular programs production in MS.NET
Recent Books
Learning An Example of student’s program with the Line
APPLICATION 2 .
The List of Standards.
“Software Engineering” Manual
APPLICATION 1 . Software Products sertifying for storing them in Repository
The Glossary of Terms.
“Software Engineering” Manual
Abbreviations&
Denotations. Assembling with Conveyor
“Software Engineering” Manual Line for Software Products Assembling from RCs and Processes
Software Product Line
Visitors: 5242 You may ask questions
Fig. 1. Main page of KNU programs factory
Programs Factory Technological Lines. The TL development is as a rule
matched with some lifecycle, e.g., implemented in the MS.NET environment with
guides, frameworks, programming languages, common libraries templates, system
tools that support new subject-oriented languages such as Domain Specific Language,
etc. For complex SP development, an assembly (compositional) line is proposed, as a
tool for composing RCs using their interfaces from various interim libraries within
development environments, as well as from RC repositories (Fig. 1). The figure
shows the main page of the web site of the factory: lines 1, 2, 3, 4 and a text-book for
e-learning fundamental aspects of software engineering [2].
The assembling conveyor has four implemented technological lines [1, 2] that,
since 2011, aid in artifacts and program development. The structure of these lines is
designed according to the technology explored by the author (Prof. E. Lavrischeva) in
1987–1991 [7, 12-16]. Simple TLs in factory are as follows:
E-learning C# in VS.NET environment (Fig. 2)
Saving components into corresponding repositories and selecting them from the
repository to meet market demands on specific SPs
258 E. Lavrischeva, A. Dzyubenko and A. Aronov
Assembling or configuring RCs into complex program structures (SP, FS)
E-learning basic knowledge on software engineering with the dedicated e-textbook
Fig. 2. Chart for components design in VS.NET environment
Web Site Lines. The depicted TL for learning C# programming is designed ac-
cording to the ISO/IEC 12207 standard of lifecycle processes while allowing for
.NET specifics as to how to perform the following design tasks:
Exploring demands on software products, singling out features and methods for
automated generation
Fixing requirements on implementations of SP functions for the domain
Specifying software elements or artifacts, documenting their passport data and
interfaces with a WSDL-like language
Storing created software artifacts and programs in the repository
The line for repository maintenance includes the mechanisms for stored RCs being
uniformly documented with their WSDL passports, as well as the tools for selecting
ready-made RCs and artifacts from the repository based on their passport data, func-
tions and relevant solution examples (see Fig. 3). This line is pertained to the proc-
esses for quality assessment of artifacts or programs created, verifying them from the
perspective of reliability and quality.
Fig. 3. Flowchart of the line for component search in repository
The line for AS development at the factory includes the tools for engineering of
specific required components using various PLs, programs and artifacts, both just
developed and selected from repository; the line also supports building multilingual
RCs and marshalling non-relevant types of data exchanged by the components [11].
The constituents of this line include standard tools for building or configuring multi-
lingual components, testing both the sample components and the links between RCs
Conception of Programs Factory for Representing and E-Learning Disciplines … 259
being composed together, as well as the tools for reliability and quality assessment
and certification of the product obtained.
The line for remote e-learning with the dedicated Software Engineering textbook
[14, 15] is now actively used in studying topics of this discipline with KNU students.
This line may also be used for independent studying in other high school institutions
where software engineering or computer science courses are established.
Design of Programs Factory. At the students’ factory, Visual Studio .NET li-
censed by KNU is used as the foundation for the programs factory; capabilities of
MS.NET platform in providing tools for multilingual AS development and support
using the components in C#, C++, Basic, etc., are utilized as well. Consequently,
third-party software developers for the factory may not be limited to the choice of a
single PL.
The factory web site is developed using PHP programming language, and, for its
external representation, HTML5, CSS3 and JavaScript. The system core is independ-
ently engineered by the authors with relying on known web frameworks [2].
5 E-learning SE Disciplines
Teaching students the aspects of software industry at Ukrainian universities is at its
initial stage. To solve several education problems, we have introduced a new ap-
proach to e-learning various aspects of SE, which assists in acquiring knowledge on
software industry.
A new concept for breaking down the software engineering disciplines (Fig. 4),
which is necessary in industrial factory production, was proposed [7, 14]. Basic goals
of software engineering disciplines are as follows:
Scientific discipline consists of the classic sciences (theory of algorithms, set the-
ory, logic theory, proofs, and so on), lifecycle standards, theory of integration, the-
ory of programming and the corresponding language tools for creating abstract
models and architectures of the specified objects, etc.
Engineering discipline is a set of technical means and methods for software devel-
opment by using standard lifecycle models; software analysis methods; require-
ment, application and domain engineering with the help of product lines; software
support, modification and adaptation to other platforms and environments
Management discipline contains the generic management theory, adapted to team-
based software development, including job schedules and their supervising, risk
management, software versioning and support
Economy discipline is a collection of the expert, qualitative and quantitative
evaluation techniques of the interim artifacts and the final result of product lines,
and the economic methods of calculating duration, size, efforts, and cost of soft-
ware development.
Product discipline consists of product lines, utilizing software resources (reusable
components, services, aspects, agents, etc.), taken from libraries and repositories; it
also contains assembling, configuring and assessing quality of software
260 E. Lavrischeva, A. Dzyubenko and A. Aronov
Fig. 4. Software engineering disciplines
In our opinion, all the SE disciplines considered above and their theoretical foun-
dations must become the independent subjects to be taught to students specializing in
the field of SE with the orientation toward the industrial production of SPs from
readymade components (reuses, services, assets, and so on) [17]. The production cy-
cle will be repeated in case of bringing changes in the product structure as it is done in
technology of continuous integration by M. Fowler and in AppFabric in MS.NET
(Fig. 5) [18].
Approach to Teaching SE. The SE educational course must include intertwined
theoretical and practical fundamental positions and achievements in software devel-
opment and integration [15, 17, 19]. The directions of the SE teaching course are as
follows:
Base concepts, principles and methods that constitute the basis of SE knowledge
and technology of programming (e.g., the five SE disciplines, life cycle, project
management, quality, configuration) and proved their productivity in practice
Mathematics of systems analysis for subject domain with the use of elements of
theory of algorithms, logic and semantics of programming for the formal design of
key notions of the domain, reflection of their communications and relations in case
of formal task of their models and SP architecture
Conception of Programs Factory for Representing and E-Learning Disciplines … 261
Fig. 5. Structure of Microsoft AppFabric
General principles and methods of designing programs, software products and
software product families using ready-made objects, components, services, aspects,
etc.
Modern applied tools for representation of software products, which are widely
used by professionals in research and development (systems analysis, decomposi-
tions, architecture, design, ООP, ontology, etc.)
Methods of measuring quality of software products
Development environments (MS.NET, IBM, CORBA, Protégé, Eclipse, etc.)
The directions of e-learning are summarized in the Software Engineering textbook.
The students learn and apply them on practice with the use of system tools, arti-
facts or programs developed by other students. Certain students’ achievements are
certified by the teacher and can be added to the repository of the factory
In terms of teaching SE, the author’s first textbook (2001), written in Ukrainian
[18], is dedicated to the foundations of SE from Curricula-2001; the second textbook
in Russian (2006) teaches the basics of Curricula-2004 [12]; the third textbook is
developed for modern approach towards teaching SE [15], including topical outlines
of some of the above-mentioned disciplines and fundamental aspects such as reliabil-
ity and quality engineering. In the new textbook, basic elements and engineering tools
are presented for development of various target SE objects and lifecycle processes,
methods for design and management of collectives of executors, quality, terms, and
262 E. Lavrischeva, A. Dzyubenko and A. Aronov
cost. The textbook on the web site describes new SE disciplines and fundamental
aspects of SE. The structure of the textbooks corresponds to the typical Curricula-
2004 program, and to the modern requirements on the subject imposed by the pro-
gram of the Ministry of Education and Science of Ukraine (2007).
6 Conclusions
The result of the authors’ work is an experimental programs factory web site, accessi-
ble on the Internet using address http://www.programsfactory.univ.kiev.ua/. This web
site is proposed for e-learning product line development at the high school institutions
on the specialties of informatics, computer sciences, information systems and tech-
nology.
The following concrete lines are established at the factory:
Production of reusable components and artifacts
Development of console applications, DLL component libraries, local Windows
applications with C# in VS.NET
Developing Java programs (a manual by I. Habibulin)
Assembling programs from RCs in MS.NET environment
E-learning software engineering with dedicated textbooks on the web site in
Ukrainian (sestudy.edu-ua.net) and in Russian (intuit.ru)
The prospects of future factory evolution are its further adjunction with new re-
sources in the field of software engineering being yet prepared by the students,
namely:
Description of the process of development of complex programs and SS using DSL
language (Eclipse-DSL, Microsoft DSL Tools)
Transformation of general data types into fundamental data types from the perspec-
tive of the standard ISO/IEC 11404-2007 generation tools
Ontological representations of new disciplines for study (e.g., computational ge-
ometry, lifecycle domains, verification)
New applied product lines for business developed with appropriate mechanisms;
SEI product lines approach, and so on
References
1. Lavrischeva, E.: Concept of Scientific Software Industry and Approach to Calculation of
Scientific Problems. Problems in Programming, 1, 3–17 (2011) (in Ukrainian)
2. Aronov, A., Dzyubenko, A.: Approach to Development of Students’ Program Factory.
Problems in Programming, 3, 42–49 (2011) (in Ukrainian)
3. Anisimov, A., Lavrischeva, E., Shevchenko, V.: On Scientific Software Industry. Techni-
cal report, Conf. Theoretical and Applied Aspects of Cybernetics. (2011) (in Ukrainian)
4. Glushkov, V., Stogniy, A., Molchanov, I.: Small Algorithmic Digital Computer. MIR, 11
(1971) (in Russian)
Conception of Programs Factory for Representing and E-Learning Disciplines … 263
5. Glushkov, V.: Basic Research and Technology Programming. Programming, 2, 3–13.
(1980) (in Russian)
6. Kapitonova, U., Letichevsky A.: Paradigms and Ideas of Academician Glushkov. Naukova
Dumka, Kiev (2003) (in Russian)
7. Lavrischeva, E.: Formation and Development of the Modular-Component Software Engi-
neering in Ukraine. V. Glushkov Institute of Cybernetics, Kiev (2008) (in Russian)
8. Czarnecki, K., Eisenecker, U.: Generative Programming: Methods, Tools, and Applica-
tions. Addison-Wesley, Boston, MA (2000)
9. Bai, Y.: Applications Interface Programming Using Multiple Languages: A Windows’
Programmer’s Guide. Prentice Hall Professional, Upper Saddle River (2003)
10. Greenfield, J., Short, K., Cook, S., Kent, S.: Software Factories: Assembling Applications
with Patterns, Models, Frameworks, and Tools. Wiley, Hoboken (2004)
11. Lavrischeva, E., Koval, G., Babenko, L., Slabospitska, O., Ignatenko, P.: New Theoretical
Foundations of Production Methods of Software Systems in Generative Programming
Context. Electronic Monograph, UK-2011, 67, VINITI RAN, Kiev, Moscow (2012) (in
Ukrainian)
12. Lavrischeva, E., Grischenko, V.: Assembly Programming. Basics of Software Industry.
2nd ed. Naukova Dumka, Kiev (2009) (in Russian)
13. Andon, P., Lavrischeva, E.: Evolution of Programs Factories in Information World. News
of NANU, 10, 15–41 (2010) (in Ukrainian)
14. Lavrischeva, E.: Classification of Software Engineering Disciplines. Cybernetics and Sys-
tems Analysis, 44(6), 791–796 (2008)
15. Lavrischeva, E.: Software Engineering. Akademperiodika, Kiev (2008) (in Ukrainian)
16. Lavrischeva, E.: Theory and Practice of Software Factories. Software–Hardware Systems.
Cybernetics and Systems Analysis, 47(6) 961–972 (2011)
17. Lavrischeva, E., Ostrovski, A., Radetskyi, I.: Approach to E-Learning Fundamental As-
pects of Software Engineering. In: Ermolayev, V. et al. (eds.) Proc. 8th Int. Conf. ICTERI-
2012, CEUR-WS, vol. 848, pp.176–187, CEUR-WS, online (2012)
18. Kolesnyk, A., Slabospitskaya, O.: Tested Approach for Variability Management Enhanc-
ing in Software Product Lines. In: Ermolayev, V. et al. (eds.) Proc. 8th Int. Conf. ICTERI-
2012, CEUR-WS, vol. 848, pp.155–162, CEUR-WS, online (2012)
19. Babenko, L., Lavrishcheva, E.: Foundations of Software Engineering. Znannya, Kiev
(2001) (in Ukrainian)
Public Information Environment of a Modern University
Natalia Morze1, Olena Kuzminska2 and Galina Protsenko3
1
Kyiv Boris Grinchenko University, Vorovskogo St. 18/2, Kyiv, Ukraine
n.morze@kmpu.edu.ua
2
National University of Life and Environmental Sciences of Ukraine,
Heroev Oborony St. 16a, Kyiv, Ukraine
olena_k@bk.ru
3 Pecherska Gymnasium № 75, A. Ivanova St. 11, Kyiv, Ukraine
galinapro@gmail.com
Abstract. Processes of society globalization and technification require changes in
training modern specialists and therefore involve changes of educational systems,
including the creation of Information environments schools. The article is de-
voted to the topics of design, development and implementation of experience of
Information environments in the educational process and scientific activities of
universities. Developed by authors model of the environment has been imple-
mented for constructing the information environment of the Kyiv Boris Grin-
chenko University and National University of Life and Environmental Sciences
of Ukraine. The article describes approaches to the training of students and teach-
ing staff of these universities for effective implementation and development of
the resources of the Information environment. Conducted monitoring of resource
usage of Information environment confirms the prospects of the authors' model
and the methodology of its introduction.
Keywords. Information environment, experience, knowledge technology, learn-
ing platform, repository
Key terms. Environment, Academia, Development, Information Communication
Technology
1 Conceptual Foundations of an Information Environment
Globalization of education leads to significant changes in the teaching systems: global-
ized learning goals, unified content and methods. New forms of technology and educa-
tion focused on the integration of information and communication technologies (ICT)
appear in the learning process. Especially significant changes occurred with the means
of learning. From the concept of "learning tools" in the traditional model of education
was made the transition to the educational environment in activity oriented teaching
practice, then to the education space in the context of person-centered, individualized
Public Information Environment of a Modern University 265
approach, and finally to the Information environment (IE), which is realized in the
process of development and ICT [1].
IE defined as a structured set of resources and technologies based on the consistent
technological and educational standards which ensures free access of persons of the
educational process to information resources, their effective communication and coop-
eration within such an environment for achieving educational goals that are known,
understandable, achievable and concrete for them preliminarily.
IE of educational institution represents (has to represent) an adaptive model of
global, national, information spaces and inherits their most characteristic functional
properties, particularly in the communicative aspect of IE is a space of co-curricular
activities based on ICT in the integration aspect provides the implementation of joint
actions by establishing appropriate rules and the adoption of regulations that means that
environment can emerge and develop only in accordance with the goals and objectives
of the above-mentioned spaces, including the regulatory framework in the field of in-
formation policy at national and international levels, the state and prospects of devel-
opment of information technology, the characteristics of learning process in educational
establishment [2].
There is a list of computer technologies in annual reports of the International Media
Consortium that will have (have) a significant impact on the organization of the educa-
tional process in the near future, namely: mobile technology and cloud computing
(2009), open content, e-books , personal web (2010 - 2011 years), semantically com-
patible programs, Smart-objects, supplemented reality (2012 - 2014 years), educational
games, sensor devices and interfaces, data visualization, training analyst (2015 - 2016
years.) [3].
In the world teaching practice Web 2.0 services are regarded as qualitatively new
means of distribution and storage of teaching materials, effective tools of educational
platforms [4, 5]. Wikis, blogs, social networks, websites and audio streaming, news
channels allow users to collaborate - sharing information data store links and multime-
dia documents, create and edit content, solve practical problems, perform educational
and research projects, etc.
That is why understanding the nature and objectives of the construction, using and
developing of information protection, a clear understanding of its structure, compo-
nents, systems development and selection of high quality resources, selection of effec-
tive service based on Web 2.0 technologies belongs to one of the main tasks of a mod-
ern University. With the current requirements of not only the operation but also the
system of educational institutions, general management principles and principles of
educational systems, as the leading principles of good design information and educa-
tional environment of the institution and of the overall architecture, should be allocated
as follows:
1. The principle of a systematic approach. This means that build model should be
based on systematic analysis of educational establishment. That means that structural
elements, internal and external communications, which will consider the educational
establishment as an open system, should be highlighted.
266 N. Morze, O. Kuzminska and G. Protsenko
2. The principle of modular structuring of information and data information. The main
purpose - to provide information and data needs in the most complete form, which
allows to characterize the state of the system and provide adequate tools for the im-
plementation of administrative functions and educational tasks.
3. Principle of modification, addition and permanent renewal. Implementation of this
principle allows for expansion, upgrading and updating of the model with additional
specific and understandable to persons indicators and measuring data. Thus, it can be
changed or adjusted in accordance with the specific educational establishment, its
traditions, mission and tasks.
4. The principle of approximation, which states that the system should be responsible
for its complexity, structure, functions, etc. to those conditions in which it operates,
and to those requirements that are set to it.
5. The principle of giving the necessary and sufficient information for the management
of educational establishment.
6. The principle of data sharing. The same data can be used by several users. In addi-
tion, each user should receive this information in an easy to view it at any time and
from any place.
2 Information Environment of a University – Concept
Realization
Information environment of the institution at the present stage should include:
Personal computing devices - a means of educational, researching and administrative
activities of the institution
Environment supporting collective and individual communication and cooperation
Open educational resources - objects of educational activities and interactions
Centralized and decentralized training platforms
Means of information security and centralized filtration incompatible with the edu-
cational process content and more
The overall architecture (organizational structure and the associated operation of
technological systems in education) is the basis of the process of creating educational
technology systems adequate to the conditions of their use, in particular, an unlimited
amount of resources that can be integrated into the educational process, a large number
of users that can use the tools and technologies of technological systems, the number of
students who may be involved in the joint solution of one educational task. Educational
environment for such systems provides by international technological standards for
interfaces, formats, communication protocols to provide mobility, interoperability,
stability, efficiency and so on.
In this approach Internet is considered as a global platform of creation and dissemi-
nation of collective knowledge. Information environment is a mean and a place for
creation, accumulation and harmonization of educational resources of efficient commu-
nication and cooperation, education and training of both students , teachers and admin-
istrators. The proposed model allows us to implement a set of technological principles
Public Information Environment of a Modern University 267
of open Information environment of the university such as adaptability, integrity, com-
plexity, interoperability.
Construction of such Information environment provides a clear projection of its ob-
jectives, functional, access channels, organization of communication of students, teach-
ers and researchers; system of continuous monitoring. The main features of the educa-
tional process in an open Information environment are:
Openness of environment - students and teachers are actively participating in the
development of educational resources and Information environment
Willingness of participants - formation of need for building individual learning tra-
jectories, positive motivation to cooperate and work in a team, willingness to dis-
seminate the results of their own educational activities in the public access
Monitoring of objects and subjects of environment - monitoring the quality of cre-
ated resources, providing access to them and their efficiency of usage, observing the
activities of the subjects of the educational process, organizing the feedback and as-
sessment
Implementation of the proposed model in a particular educational institution in-
volves the selection of platforms (Fig. 1) and resources:
Scientific articles of educational-research and Masters members of National Univer-
sity of Life and Environmental Sciences of Ukraine
http://elibrary.nubip.edu.ua
Abstracts of theses defended in National University of Life and Environmental Sci-
ences of Ukraine http://elibrary.nubip.edu.ua
Conference proceedings of National University of Life and Environmental Sciences
of Ukraine http://elibrary.nubip.edu.ua
Works of magisters of National University of Life and Environmental Sciences of
Ukraine http://elibrary.nubip.edu.ua
Training materials to support the educational process of National University of Life
and Environmental Sciences of Ukraine http://elibrary.nubip.edu.ua
e-Learning of National University of Life and Environmental Sciences of Ukraine
http://moodle.nauu.kiev.ua
Distance learning http://agrowiki.nubip.edu.ua
Harmonized (supplemented by expert comments of National University of Life and
Environmental Sciences of Ukraine and links to internal and external resources)
standards http://agrowiki.nubip.edu.ua
Regulations of National University of Life and Environmental Sciences of Ukraine
http://elibrary.nubip.edu.ua
Thematic practice-oriented information articles http://agroua.net
268 N. Morze, O. Kuzminska and G. Protsenko
Fig. 1. Open Information Environment of National University of Life and Environmental Sci-
ences of Ukraine
Effective usage of resources of the Information environment is largely dependent on
the willingness of teaching staff to implement innovative pedagogical and information
technologies and work with students in IE.
Question of training faculty can be solved on the basis of properly designed training
system focused not so much on the study of specific technologies as on:
Formation among the teachers methodical approach to the selection and usage in
their professional activities IE resources to achieve educationally meaningful results
in the context of ensuring the availability of educational materials, improving the
quality and effectiveness of the educational process
Developing skills of the educational process with the use of IE resources and manag-
ing innovative educational projects
Logical design and creation of ICT-oriented learning tools
While building a training system for teaching staff to use IE resources in educational
activities it is necessary to take into account the necessity:
Modular structuring of content that reflects the technological and didactic possibili-
ties of the usage of specific resources IE
Malancing and harmonizing individual content modules training program for teach-
ers
In the process of training teachers act as students (http://lilia.moyblog.net/category/)
and this allows to simulate educational situations, identify problems and use of IE com-
ponents to create training courses of new sample
Public Information Environment of a Modern University 269
(http://moodle.kmpu.edu.ua/dn/course/view.php?id=144¬ifyeditingon=1). Topics of
workshops:
Institutional repository and its role in the creation of electronic educational and re-
searching environment of the University
Platform of e-learning Moodle
Safe work in the Internet
Creating of modern ontology on a base of wiki-portals
Role of ICT in usage of formative assessment
Usage of Web 2.0 for the students individual work
Blogs and their usage patterns in the educational process
Podcasts usage in educational process
Role of ICT in the organization of cooperation and communication
Creation of Information environment will provide flexible formation of educational
and methodical complexes according to the different models of learning, make teaching
materials cheaper and more accessible, improve learning efficiency by providing shar-
ing of experiences and a variety of educational materials between students and teachers.
Organization of educational and research activities in the Informational environment
determine what skills a student should possess, namely:
Access to information data and resources – the ability to search, collect and store
information data
Management of information data resources – the ability to choose existing resources
for categorizing and structuring information data
Critical evaluation of information data and resources – the ability to make judgments
about the data quality, importance, usefulness or effectiveness as well as the reliabil-
ity, specific and address orientation
Creation of information data - the ability to interpret and present data, generate data
and knowledge
Exchange of information data – the ability to transmit information data by means of
information environment in a proper way
Acquiring of the mentioned skills and abilities is developed in the process of inde-
pendent activity of students, such as the preparation and defense of Master’s Thesis.
The wiki-portal EcoAgroWiki is chosen as a technological platform of the informa-
tional support for Master’s educational activity. Considering the functionality of wiki
technology article authors managed to organize a community of masters, teachers, aca-
demic advisors practice (see Table 1). In order to simplify page layout on the EcoA-
groWiki portal according to the standard IMS ePortfolio Information Model
(http://www.imsglobal.org/ep/epv1p0/imsep_bestv1p0.html), it was designed templates
of teachers and students portfolio. The results of network cooperation of its members is
collections of useful links, digital electronic resources, target selection and description
of the use of modern software tools, organization of project activities and collaboration,
system of effective communication, consultation and expert evaluation.
270 N. Morze, O. Kuzminska and G. Protsenko
Table 1. Subject of seminars series “Presenting of Masters’ scientific researches’ results using
ICT”
Annotation Resources
Topic 1. Electronic publishing
Scientometrical bases (ЕBSCO). http://web.ebscohost.com/ehos
FAO- resources: AGORA. t/search
http://www.fao.org/index_en.h
tm
Topic 2. Institutional repository
Institutional repository of NUBiP of Ukraine. http://elibrary.nubip.edu.ua
OAI harvester of Ukraine. http://oai.org.ua
International repository.
http://arxiv.org/
Topic 3. Bibliography
Resources description variants. http://agrowiki.nubip.edu.ua/
Personal bibliographic managers. wiki/index.php/Zotero
Topic 4. Research results publication in the Internet
Google documents, blogs, forums, conferences. http://apctt.blogspot.com
Topic 5. Cooperation organization
Google groups.Wiki-portal. http://agrowiki.nubip.edu.ua
University portal. LMS. http://it.nubip.edu.ua/
http://nubip.edu.ua
3 Experience of Resources of the Open Information
Environment’s Usage by Students and Teachers
The real impact of the Open Informational environment into educational activities or-
ganization of the university was determined by on-line poll on wiki portal EcoA-
groWiki and on-site discussions of the educational process. Master’s programme stu-
dents of the NUBiP of Ukraine faculty of Computer Sciences and of the faculty of
Ecology took part in the poll.
Before series of seminars authors examined the attitude of students towards imple-
mentation of the Information environment resources in teaching and research activities.
The vast majority of students (189 of 200 respondents) believe that creation of Open
Information environment significantly enhance the information support of education
activities. However, 60% masters in computer sciences (hereinafter referred to as
Group 1) and 20% of ecologists (hereinafter referred to as Group 2) consider effective
the use of e-learning courses developed on the platform LMS Moodle, 50% of Group 1
and 30% of Group 2 use the institutional repository for viewing topics and presenta-
tions of Masters’ Thesis of previous years. The question with which search engines are
students required information, including these with a scientific character, the over-
whelming number of respondents named Google, and to save search results 70% of
Group 1 and 50% of Group 2 use personal folders and file cards on personal computers.
As of communication between students, most of them called social networks, and as of
Public Information Environment of a Modern University 271
communication with teachers (Academic Advisors) – personal correspondence via e-
mail.
There are three the most important students’ opinion, directions of the Information
environment use. Students argued their choices from the position of information liter-
acy [6]. 80% of Group 2 and 60% of Group 1 noted acquiring skills for analyzing the
obtained data, sites, resources actively and productively; planning and managing their
studying; establishing effective communication and cooperation; solving problems
together, selecting the most effective resources and technologies to solve specific tasks.
Interviewed teachers (42 educational research worker of NUBiP of Ukraine) noted
the stiffening of Masters’ Thesis, especially in a part of analysis of research problem
development, usage of modern resources, in particular materials of open scientific jour-
nals and bibliography description. In addition, Academic Advisors, who have joined the
experiment, noted the increase of their own computer literacy through the usage of
scientific communication means and cooperation in the process of joined work together
with students. And creation of teachers’ portfolio also make for promoting University
activities, searching for partners in joint projects etc.
4 Conclusions
Experience of Information environment creation and usage of its resources in the Kyiv
Borys Hrinchenko University and NUBiP of Ukraine suggests the following arguments
in favor of the proposed model: individualization and personalization of the academic
activity, quality, flexibility, ability to meet the educational requirements, timeliness,
self and mutual control, cooperation. Open Information environment have to be built as
a system of functionally and structurally interconnected information and technological
elements, skillful usage of which allows a teacher in practice solve didactic tasks on the
technological basis with a guaranteed quality in the age of education informatization.
Information environment creation at the level of educational institution leads to that
educational materials and services will be available to every subject of the educational
process. As a result, conditions for equal access to a quality education will be formed –
an opportunity for everyone to learn at any place and at any time become a reality.
Under these conditions, the Information environment is potentially unlimited as to the
available quantitative and qualitative number of educational resources (can be used in
the education process), number of users (can use its resources and technologies) and
number of subjects of educational activities that can work together for solving educa-
tional tasks.
References
1. Manako, А. F.: Evolyutsiya ta Konvergentsiya Informatsiynyh Tehnologiy Pidtrymky Os-
vity ta Navchannya. In: Proc. ITEA-2011, pp. 3–19, IRTC, Kyiv (2011) (in Ukrainian)
2. Bykov, V. Yu.: Avtomatyzovani Informatsiyni Systemy Yedynoho Informatsiynogo Pros-
tory Osvity i Nauky. Zbirnyk Naukovyh Prats Umanskoho Derzhavnoho Pedahohichnoho
Universytetu im. Pavla Tychyny, Ch. 2, 47–56 (2008) (in Ukrainian)
272 N. Morze, O. Kuzminska and G. Protsenko
3. The Horizon Report: 2011 K-12 Edition, New Media Consortium,
http://www.nmc.org/pdf/2011-Horizon-Report-K12.pdf (2011)
4. Blees, I.: Web 2.0 Learning Environment: Concept, Implementation, Evaluation. eLearning
Papers, 15, 18, http://www.elearningeuropa.info/en/article/Web-2.0-Learning-Environment
%3A-Concept%2C-Implementation%2C-Evaluation (2009)
5. Malinka, I.: Involving Students in Managing their Own Learning. eLearning Papers, 21, 13,
http://www.elearningeuropa.info/en/article/Involving-students-in-managing-their-own-
learning (2010)
6. Information Literacy at Otterbein College. Ottebein University, http://www.otterbein.edu/
resources/library/information_literacy/index.htm
Designing Massive Open Online Courses
Vladimir Kukharenko 1
1
National Technical University “Kharkiv Polytechnic Institute”,
Frunze Street 21, 61002 Kharkiv Ukraine
kukharenkovn@gmail.com
Abstract. Connective massive open online courses (MOOC) for teachers from
Ukraine and Russia were conducted in 2011-2013. They were: Strategy of Dis-
tance Learning in the Organization, Social Services in Distance Learning, Dis-
tance Learning from A to Z, Designing Online Courses. The accumulated ex-
perience allowed to develop recommendations for each ADDIE step of MOOC
designing for the Russian-speaking audience.
Keywords. Connectivism, massive open online course, personal learning envi-
ronment, ADDIE
Key terms. Competence, Didactics, TeachingMethodology, TeachingProcess,
ICTEnvironment
1 Introduction
The term “massive open online course” (MOOC) was introduced by Dave Cormier in
George Siemens’s distance course “Connectivism and Connective Knowledge” in
2008 and 2010 (http://connect.downes.ca/). This course was devoted to the problems
of a new learning theory – connectivism, according to which learning is the process of
creating a network (more than 2200 people studied). Units of such a network are ex-
ternal entities (people, organizations, libraries, websites, books, journals, databases, or
any other sources of information). The act of learning implies the creation of the ex-
ternal network units. Over the last few years several dozens of open online courses
have been conducted. Those courses are based on the new approach named “connec-
tivism” and therefore abbreviated as cMOOC. cMOOC is characterized [1] by a
structured network, the use of daily bulletin, a big amount of information material, the
social approach to teaching. cMOOC enables people to cluster around the central
core.
In cMOOC the teacher plays a lot of roles [2]: he is an amplifier, tutor, he directs
and socially manages creation of meanings, he filters, models and is always present.
The student’s success in cMOOC is provided by his ability to navigate the net-
work, the formed personal learning environment (PLE) and personal learning network
274 V. Kukharenko
(PLN) as well as his personal goals. Personality development and personal learning
play a central part in in cMOOC [1].
Experts believe that cMOOC [3] is suitable for effective independent learners, who
have learned to select content. The reduced participation can be neither good nor bad.
MOOC can be most effective as a form of continuous education and perfecting skills.
2 Analysis of cMOOC of National Technical University
“Kharkiv Polytechnic Institute”
The courses are free of the traditional content. For each week is given an extended
abstract and a set of links to various materials on the subject. A brief analysis of the
topic is on the webinar, position of experts considered the on the guest webinar.
Before the start of the course students are given instructions about the features of a
connective course. Notes the desirability of selectively view recommended materials,
preparation of remixes, posting material online and active participation in the discus-
sion.
2.1 Distance Course "Strategy of e-learning Development in the
Organization"
The open distance course "Strategy of e-learning development in the organization"
lasted 6 weeks in February - April 2011 [4].
The main objective of this course was to show the possible uses of e-learning in the
organization and to assist in the development of strategy of learning that takes into
account the overall strategy of the organization; to learn the designing of the learning
process in the open distance course, to assess readiness of the Russian-speaking audi-
ence to study in the new environment.
The target audience comprises teachers, post-graduate students, heads of educa-
tional departments from various organizations. The participants are expected to have
skills of working in Internet, social networks and means of web communication (syn-
chronous and asynchronous) for realizing communication, collaboration and exchange
of information.
45 persons were registered for the course, pages of the course were visited by more
than100 people, 12 people passed the final survey. The questionnaire was completed
by the same number of participants from the academic and corporate sectors with
experience in educational work of more than 5 years (83%) and the experience of
distance learning of more than 3 years (67%).
3 participants couldn’t formulate their goal of taking this course, the rest of the par-
ticipants (8 people) had the goal of getting acquainted with the features of open dis-
tance learning course and new social services.
The main activity of the participants could be traced by a mailing list (about 200
messages, half of them were sent during the first two weeks of classes, then the activ-
ity decreased); in Moodle only the forum of dating worked, all other invitations to
discussions were not supported. Based on the materials of the course, two participants
Designing Massive Open Online Courses 275
created blogs. Generally, the group worked passively – its members read the proposed
materials, did not disclose their sources, did not give their points of view on specific
topics of the course.
There have been conducted six weekly, introductory and final webinars which
were attended by about 10 participants each. All webinars were conducted in the envi-
ronment of WIZIQ, one of them – in a virtual world “VAcademy”, where the partici-
pants got acquainted with virtual environment possibilities. Besides, 3 guest webinars
also took place.
The experience of giving such course shows that the open course for CIS audience
is a new and not always obvious concept, the great amount of instructional material
and absence of clearly stated ideas cause great difficulties for participants. The limited
set of social services, disrespect and misunderstanding of Twitter summons problems
when tracing tutors’ and their colleagues’ work. Apart from it, the author thinks that
the topic of the course was rather challenging for the learners. The fact that the per-
sonal instructional environment is not formed was the reason of problems arising
during the learning process [4].
2.2 Distance Course “Social Service in Distance Teaching”
An open distance course “Social Services in Distance Course” [5] was held from May
23 to July 3, 2011 [6]. For this course Wiki, Mailing List, Twitter, DIIGO, learners’
blogs and aggregator netvibes.com were used.
This study examined the hypothesis that introductory webinars on forming per-
sonal learning environment (PLE) and the Workshop will increase the activity of
students and material will be drafted for discussion by the full-time session. Besides,
the introductory webinars were supposed to be held by a group of tutors.
Since the beginning of subscribing to the course active attendance of the course
and surfing the pages started. By the moment classes began 43 people had been sub-
scribed. At the beginning of the course we could observe the maximum of visits and
browsing. Then the attendance was gradually decreasing. It should be noted that the
number of visitors (Ukraine – 64%, Russia – 23%, the USA – 9%, Belorussia – 4%)
was twice as big as the number of those registered.
30 people passed entrance questioning, 83% of them are university lecturers, the
others are from the corporate sector; 74% of people have teaching experience of more
than 5 years, 80% are experienced in making distance courses; 57% are experienced
in tutor’s work; 37% use mobile phones to access the Internet.
It is essential that personal learning environment of the course’s participants is
poorly formed, only 10-20% of participants use most services which is not helpful in
communication.
Most of the participants stated the aim of the training as a necessity to master the
new social services and outline the new approaches to distance training on their base.
On completing the course 10 participants took part in the questioning. On average
the learners spent 6-8 hours a week.
276 V. Kukharenko
2.3 Distance Course “Distance Course from A to Z”
The open distance course “Distance Course from A to Z” includes two parts. The first
part which was held from December 5, 2011 to January 22, 2012 was devoted to the
tendencies of creating the system of distance training at the current stage of Internet
development. In the second part - the distance course design methods and the distance
learning process.
The main objective of the course is to analyze the level of distance training devel-
opment in Ukraine on the base of webinars held in May-October 2011 [7], consider
the tendencies of distance training development abroad and outline the requirements
to the modern system of distance training.
The course is based on the blogs and articles published 2-3 months before the
course started. Such references show modern tendencies in the distance learning sys-
tem mainly abroad. To shape the references Twitter was used.
Before the course started the following webinars were held: “Social Services in
Teaching Process”, “Twitter”, “Personal Learning Environment”. Their task was to
help the learners acquire the skills of using social services during the course.
The total number of those registered in the course is 31. According to Google
Analitics the course was attended by 430 people from 29 countries of the world, 117
cities. Among them: from Ukraine – 76%, Russia – 13%, Belorussia – 5%. On aver-
age there were 30 people a day. Surfing the weekly pages ranges from 550 to 200.
During the course 12 people wrote 68 blogs, 85 messages were left in the course.
More than 80% of participants have teaching experience of more than 5 years, 68% of
them – more than 10 years. The experience in distance course usage is from 1 to 10
years.
22 participants answered the question of the final questionnaire. They are experi-
enced teachers with 10-year experience of pedagogical work and 5-year experience of
distance training, half of them worked near 3 hours in the course (3 persons worked
more than 8 hours). During the course most of the participants stated their aims,
which can be generally outlined as follows: to obtain and systematize the ideas of
modern distance training and other universities’ experience.
Answering the question about the novelty of the distance course all participants
mentioned such services as Twitter, DIIGO, Creative Commons licences, personal
learning environment, open learning resources and open distance courses, new ap-
proaches in distance education.
The participants of the course liked the system approach to the problem, the format
of the course, unobtrusive and not strict management, great number of various infor-
mation resources, possibility to work with information and new services in a suitable
time, openness of communication, active information and experience sharing, the
potential efficiency of the course, as under certain conditions the mechanism of self-
reproduction of the course can be launched, that is, it will start working without visi-
ble participation of the organizers.
The learners noted the challenges of the open distance course, such as mastering
new tools, personal aim of the study (dynamic and not always concrete), work with
great volume of unstructured information in foreign languages.
Designing Massive Open Online Courses 277
The participants’ work and their impressions of the training process of Mass Open
Distance Course were considered at the Xth International Seminar “Modern Peda-
gogical Technologies in Education” which took place on January 31 – February 2,
2012 [8]. One of the suggestions was to organize the groups by interests.
It should be noted that the attendance of courses did not stop after they were offi-
cially closed in April 2012.
On the whole during the year the course was attended by 2492 users, most of them
live in Ukraine (Kharkiv, Kyiv, Odessa) – 60%, Russia (Yaroslavl, Moscow, Yekater-
inburg) – 20%, and Kazakhstan (3%).
2.4 Distance Course “”E-learning Design”
In 2012-2013 academic year the Research Laboratory of Distance Learning made an
attempt to hold a combined MOOC course "E-learning Design”, which consists of a
course for beginners (xMOOC constructivism approach) and that for heads of dis-
tance learning centers of an organization (cMOOC connectivism approach). The free
program of the course includes three modules: “Basics of Distant Learning" (6- week
long), "Technology of the Development of Distance Learning Course" (12-week long)
and "Practicum for Tutors" (6- week long).
The objective of the course is to improve and standardize the level of teacher train-
ing in distant learning for colleges and universities in Ukraine.
Experienced professionals (managers) of distance learning take the course, built on
the connectivism principle, in order to organize education for distance learning course
developers in their organizations. The educational process of the course encourages
the exchange of experiences among the students, improving the quality of the course.
Students plan the training for distant-learning course developers either independently
or making use of the proposed distance course for beginners. The overall goal of this
course is to improve the efficiency of training for distance learning course developers.
To participate in this course, you should be able to use social services (twitter, RSS,
netvibes, paper.li, scoop.it and others) and log in the open course [9].
The second course, built on the principles of constructivism, is where beginners
learn to create their own distance courses. The tutor provides general management of
the educational process, the leaders of distance learning centers may act as local tutors
for their teachers. If desired, these students can take part in the first distance learning
courses as well.
As many as 90 individuals from the academic (91%) and corporate (9%) sectors,
with distance learning (45%) and tutor (46%) experiences logged in the course. Most
of the students (64%) have professional experience of over 5 years. Most students
(43%) spent over 5 hours each week working on the course, 3-5 hours per week
(31%), with the rest spending less than three hours weekly (25%).
The intention of the students involved were focused mainly on observation of
course events (80%), participation in discussions (75%), establishing new contacts
(72%), creating a distance course (68%).
278 V. Kukharenko
After the first distance-learning course "Basics of Distance Learning" the open
connective part of the Wiki course was closed because of student low activeness
while the course participants were active in making their assignments.
The second part of the course "The Development of Distance Learning Course
Technologies" envisaged the involvement of groups of distance courses developers
from various universities together with their instructors. However, only the teaching
staff of Kharkiv National Pharmaceutical University accepted the invitation. During
the educational process, they were very active in various forums, helped each other
and created distance learning courses.
Among 61% of the logged-in students involved in the training, 40% worked ac-
tively, yet only 11% (9 people) complied with the course program.
Apart from creating distance courses, the students filled out workbooks (5 tasks,
performed by 29…12 participants), questionnaires (6 tasks, done by 34…12 partici-
pants), did two tests (32 and 27 participants), and discussed various issues in 10 fo-
rums. More than 5 hours per week were spent to work on the course by 43% of stu-
dents, from 3 to 5 hours a week – by 31% and 25% - less than 3 hours a week.
3 cMOOC Design
The courses held allowed recommendations on the ways to design cMOOC for Rus-
sian-speaking audiences. It should be noted that currently the MOOC design is being
focused on technical aspects, i.e. the choice of content media placement, communica-
tion, aggregators and other services. Stephen Downes [10] has given some recom-
mendations for cMOOC design.
MOOC design is supposed to use the ADDIE (Analyzing, Designing, Developing,
Implementing, Evaluating). This approach is an adaptation of the design methods of
technical facilities for the pedagogy.
3.1 Analysis
cMOOC features uncertain audience and a variety of learners’ purposes. Therefore
themes of a distance course are determined by the author of the course, or, which is
preferable, by the author and his/her team.
The themes must be relevant, contemporary, and a lot of unstructured information
should be generated in the selected area. A theme is the best choice if it is the object
of author’s research. In this case, the author can formulate his/her research aims, this
leading to the program of the course.
The main problem at this stage is a small audience and its inactiveness, poor PLE.
Most often, students are targeted at getting acquainted with some information and
establishing new contacts. A course participant must have essential implicit knowl-
edge on the theme of the course, technical skills to work with social services and time
management tools.
Designing Massive Open Online Courses 279
3.2 Design
After completing the course program, the duration of the course should be deter-
mined. The course duration is desirable not to exceed the time period of 6-8 weeks as
students find it difficult to concentrate for longer periods. The next step is to draw up
draft descriptions for sections of the course.
A most critical step is to select informative materials. To do that, the author of the
course should possess the content curator skills. In this case, a vast number of links on
Twitter to various information resources treated as blogs, e-magazines like scoop.it
and much more are made available by the time the course is ready, which is among
the content curator’s function.
The most recent links to informative materials should be selected; and it is desir-
able to choose the number of links on the topic that are above Dunbar’s number in
order to organize selective information processing for students. Dunbar's number is a
cognitive limit to the number of people with whom one can maintain steady social
relationships (100 - 230, 150 selected) [11]. With a small number of links, students
psychologically tend to seek reading all materials.
After selecting the informative materials it is advantageous to write a brief abstract
for each link, e.g. to use the tools cruxlight.com, and sort them out according to areas
so as to make students’ work easier.
3.3 Development
Now comes the time for a summary for all practical hours, with problem-setting and
specifying various tasks for students. A list of recommended student activities is de-
sirable, e.g. to make about 10 retweets, write 2-3 blogs with a review of some sources
etc. The course should contain a variety of recommendations and references for those
who have little experience of using social services.
It is also necessary to develop entry and graduation questionnaires. Weekly ques-
tionnaires are recommended to include questions:
Which materials were of interest? Why (not)?
What was new and interesting in the webinar?
Which can be considered an alternative blog?
Students can choose between filling out the questionnaires and writing a blog (re-
mix, reflection).
The next step is to select guest lecturers for each week. They can be author’s col-
leagues or involved students who are leading experts in some aspects of the course.
3.4 Implementation
MOOC should be conducted by a team of tutors. Their roles may be different. For
example, they can share functions determined by the main author of the course. The
tutors can be independent, each of them preparing their own materials for the course
and giving his/her opinions during webinars without prior consultations. The most
280 V. Kukharenko
important thing is that teams of tutors work very actively thus setting the work pace
for the audience.
Before the MOOC starts, it is advisable to hold 1-2 webinars telling about the tools
employed, especially twitter, evernote, mailing lists, blogs, and use of translators.
This could be helpful for students to get prepared for the educational process.
The process of designing a distance course is an iterative process of creating a fun-
damentally new service. It requires studying other authors’ experiences, new social
services, continuous work with new information related to course themes.
4 Resume
The experience shows that it is very difficult to introduce cMOOCs in education and
distance learning in the CIS. This could be due to the choice of themes for courses,
low activeness of the pedagogical community, a small number of social services used.
For example, Twitter is not very popular in the educational community. At the same
time, four courses held led to formation of a community of as many as 30...40 mem-
bers who actively participate in all cMOOCs organized not only in the CIS.
Low activeness of cMOOC participants does not allow implementing the principles
of connectivism. Therefore, open courses are first needed on the use of social services
in education, followed by training content curators. Furthermore, specific features of
CIS audience must be considered when designing cMOOCs.
References
1. Downes, S.: Education as Platform: The MOOC Experience and What we can Do to Make
it Better. March 12, http://halfanhour.blogspot.com/2012/03/education-as-platform-mooc-
experience.html?spref=tw (2012)
2. Bosman, J.: Teacher Roles and MOOC, http://moocblogcalendar.wordpress.com/2012/
03/19/change11-teacher-roles-and-mooc/ (2012)
3. Stevenson, D.: MOOC- The Recent Discussions about MOOCs. http://learning-
aworkinprogress.blogspot.com/2012/03/change11-mooc-recent-discussions-about.html
(2012)
4. Kukharenko, V.N.: Innovation in e-Learning: Massive Open Online Course. High Educa-
tion in Russia, 10, 93–98 (2011) (in Russian)
5. Distance Course: Social Services in Distance Learning. http://el-ukraine.wikispaces.com/
(in Russian)
6. Kukharenko, V.M.: Learning Process in Massive Open Online Course. Management
Theory and Practice in Social Systems, 1, 40–50 (2012) (in Ukrainian)
7. Webinar Records of Chief of Center of Distance Learning Ukraine Universities.
http://dl.khadi.kharkov.ua/mod/resource/view.php?id=8404 (in Ukrainian)
8. Х Workshop Materials, http://dl.kharkiv.edu/mod/resource/view.php?id=11229 (in Ukrain-
ian)
9. Distance Course: Design e-Learning. http://de-l.wikispaces.com/ (in Russian)
10. Downes, S.: Creating the Connectivist Course. http://moocblogcalendar.wordpress.com/
2012/01/03/creating-the-connectivist-course/ (2012)
11. Wikipedia: http://en.wikipedia.org/wiki/Dunbar 's_number (2012)
The Role of Informatization in the Change of Higher
School Tasks: the Impact on the Professional Teacher
Competences
Dmitry Bodnenko1
1
BGKU, Borys Grinchenko Kyiv University, 18/2 Vorovskogo st. 04053 Kyiv, Ukraine
bodnenko@kmpu.edu.ua
Abstract. Last decade is characterized by the tendencies of the modern higher
school development. This is based on the informatization of education. In this
work, basing on didactics of higher school pedagogy a system of psychological
and pedagogical characteristics for higher school teachers was created. Psycho-
logical, educational requirements for professional competence of university
teachers are based on components of pedagogical skills of teachers of higher
education institutions and absorb almost all of its functions, duties and skills,
but increased usage of ICT determines the specification of the classification
system of psychological and educational requirements for information and
communication competence of the university teacher.
Keywords. TeachingMethodology, competence, distance education, technolo-
gies of distance education, tutor, listener, course of distance education, informa-
tive resource, network services
Key terms. Didactics, CompetenceFormationProcess, InformationCommunica-
tionTechnology, TeachingMethodology
1 Introduction
In the conditions of the pedagogical paradigm updating, emergence and distribution
of network technologies, and consequently, enrichment of personality aspects of
modern teacher preparation in higher school, a large value is acquired by interpreta-
tion of the teacher's professional competence concept.
The problem explored by us in this article is concerned with how the informatiza-
tion of education influences the requirements for professional competence of teaching
at universities.
During the recent years, in Ukraine ICT development has been defined as one of
the priorities. In 2007 the Parliament passed the Law "About the Basic Principles of
information Society’s development in Ukraine in 2007-2015" (№ 537-V of
9.01.2007), pursuant to the Action Plan has also been adopted. Both documents were
designed to promote the development of information society and the introduction of
282 D. Bodnenko
information technology as apriority direction of the state policy. The need for further
development and implementation of ICT is confirmed, also, by a number of national
documents, such as the Draft of National Education Strategy in Ukraine for 2012-
2021 years CMU (Meeting of 11.09.2012), the Laws of Ukraine "About the concep-
tion of the national programme of informatization" (№ 75/98-VR of 04.02.1998) with
amendments introduced according to the laws N 3421-IV (3421-15) of 09.02.2006,
VVR, 2006, N 22, article 199, N 3610-VI (3610-17) of 07.07.2011).
Unfortunately, these intentions, mostly in the present time, are remaining on an
embryonic stage, that is reflected in the low Ukrainian ratings of competitiveness and
network readiness.
Implementation of informatization at the universities hasn’t been explored fully
yet. This topic opens up many opportunities for further studying and investigation:
implementation of educational programs, websites, the use of network services, an
invention of new forms, ways, exploring of university teachers potential at the new
open education environments. Writing this article we set the following goals: explor-
ing of the university teachers’ professional competence; the definition and justifica-
tion of psychological and educational requirements for information and communica-
tion competency of University teacher.
2 The Main Part of Our Research
In recent years, the term "tutor" became well-known, it gained a considerable popular-
ity in higher education, giving ground in frequency of use only to the term "teacher".
These notions are almost synonymous, which could cause some dubieties about the
necessity of introducing the new foreign term. But, in fact, the token tutor expands the
notion teacher, especially in the context of the gradual integration of Ukraine into the
European educational space.
S. Goncharenko [4] notes that the term tutor (English tutor from Latin tueor, to ob-
serve, to care) – is a teacher-mentor in British "public schools", high forms of gram-
mar schools and teachers colleges [4].
In modern educational paradigm, regardless of the learning forms, the principle
about the student’s importance was established, that fact illustrates a student as a
dominant. According to this fact, teachers act as assistants, as friend, as mentor, who
supports students in getting education. With the rapid growth of informational and
communicational component of the educational process, particularly in the context of
the implementation of distance studying, it is difficult to imagine a teacher, who just
transmits the information to the listener, even if it is a video lecture. Teachers need to
be a coordinator, facilitator, who "synthesizes and accompanying student’s resources"
[6], which has much bigger freedom of the choice (educational content, time, place,
methods of education), compared with the traditional student.
We support the aspects of the network training’s specific which is - pointed out in
V.M. Kukharenko’s and V.Y. Bykov’s studios [5], which claim that the teacher-tutor
should focus on his practice, doing his classes for the listener of the e-learning course.
The Role of Informatization in the Change of Higher School Tasks … 283
However, working in the field of distance studying, the teacher communicates with
the diverse contingent.
Each person is a personality, according to some needs, abilities and opportunities.
That is why the teacher’s task is to choose the best way of the studying process. They
have to coordinate their activities (growth, educational content, methods, tools) with
options of the audience potential or take responsibility for themselves, creating a new
pace of studying. Also they should to select, a group of listeners who exactly over-
take the course (it is not the fact that all students will be able to take this rate).
As the research is based on the condition, that the teachers of the higher educa-
tional institutions are studied for the implementation of the e-studying course, so we
agree with S.S. Vitvytskaya’s supposition [3] that teachers have to learn to perform
the following functions: organizational (the head, the leader in the maze of the knowl-
edge and skills); informational (the carrier of the last information); transformational
(transformation the socially meaningful content of the knowledge to the act of the
personal knowledge); orientational and regulative (the teacher’s structure of the
knowledge determines the structure of the student knowledge); catalytic (transforma-
tion the object of the education into the subject).
The teacher, who is not at the top level mastering pedagogy of higher education, it
is difficult, in our view, to prepare for the implementation of distance learning in
higher education.
According to A.I. Kuzmynskyi [7], the higher school teachers have some compe-
tence (the high professional competence, pedagogical competence, social and eco-
nomic competences, communicative competence), have a high level of general culture
and, also we can highlight especially important, in our mind, for the introduction of
ICT competencies the functional responsibilities the universities’ teachers, in particu-
lar. All these ingredients along with the pedagogical skills (except some components
of the educational technology), outlined in the A.I. Kuzminskoho’s works [7], form
the basis of psycho-pedagogical portrait of the teacher, and transforming the system
MDs we will get in the future - a distance studying tutor or teacher, who has informa-
tion and communication competences, who is ready to work upon condition of a new
paradigm of education. In particular the pedagogical skills include: moral and spiri-
tual qualities, professional knowledge, social and pedagogical qualities, psychological
and pedagogical skills, pedagogical technique.
Russian authors, as G. Adrionova [1], M.V. Vislobokova [11], V.P. Verzhbytskyy
[2] say that the traditional teacher and teacher of the e-studying - are mutually differ-
ent personalities with the different characteristics. We disagree with this opinion,
because with the help of the teaching skills of higher school teachers (as one of the
key characteristics of the higher school teacher) it was possible to form the teacher -
oriented on the using of new teaching technologies, including ICT. The fact that many
characteristics are really opposite, because they depend on the tasks set for the teacher
and the student.
The most part of the scholars and teachers asserts [8, 9]: teacher – is a basis of the
educational process, the most important component, that organizes the e-studying
process and ensures its quality. V.M. Kukharenko [5] emphasizes the role of the tutor
284 D. Bodnenko
in the e-studying system: "Any course requires the tutor, but a good course - requires
skillful tutor".
Distance studying technologies that are introduced at the time of the rapid devel-
opment of interactive technologies, absorb current dominant of educational paradigm:
the activities of the teacher-tutor designed to organizing, promoting and supporting of
the students’ independent learning activities at the distance learning courses, which
has the development of creativity as a basis, developing of searching abilities, analyz-
ing and organizing information and rendition on the basis of the the right decisions’
findings. In our opinion, to develop the creative abilities of the listeners can only the
teacher, who also is a creative person.
In the context of the above statement is A.M. Eagle’s saying [10] that the tutor – is
a teacher of the high level, who is able to interact with the audience, producing new
relaxed and fun lessons for all the. The researcher says that the tutor doesn’t have to
teach the audience, he has to maintain him as long as the student takes sufficient in-
dependence and competence.
The proof of this distribution is the work of the authors from the Problem Labora-
tory NAM NTU "KPI" [5], where the experience of the foreign researchers [12,13] is
adapted to the Ukrainian Distance Learning System and where is outlined some tu-
tor’s responsibilities, according to two stages (development of the course and organi-
zation of the educational process). We have to note that in this case the tutor can be a
user of already created e-studying course.
So systematizing the lined material, taking it as a basis, based on the generally ac-
cepted didactic principles, which are clearly defined in the A.I. Kuzminskiy’s work
[7], form the system of psychological and educational requirements for informational
and communicational competences of the university’s teacher.
Informational and communicational skills include the following components:
To have at least one information-educational environment
To know the range of services provided by the environment and technology of the
handling these services;
To know the basic principles of the telecommunication systems
To know the specifics of the webinar, audio, video- teleconferencing, chats and
forums;
To know the rules of conducting (etiquette) during the interactive dialogue
To know the specifics of working within formational resources (databases, infor-
mation services)
To be able to use the communication capabilities of computer’ networks to organ-
ize fruitful communication between the participants of the educational process
To make the organization and conducting of the telecommunication project
To own and use network services in a professional activity
Didactic skills:
To create and shape the course material for students of e-learning courses with the
optimal (understandable, accessible, scientific) laying out the information to ensure
personal, effective and independent from the listener’s time and his working place
To implement psycho-pedagogical monitoring (previous, current, interim, final)
The Role of Informatization in the Change of Higher School Tasks … 285
To manage the independent educational and cognitive activity of the students, to
develop intellectual capabilities and to form the motives of the education
To teach students the efficient and effective methods of independent activity in the
educational process
Constructive skills:
To integrate and combine full-time, part-time, external and distance learning
To create e-learning and /or correct an existing course, according to the educational
process requirements
To adopt the effective types and forms of participants’ activity of the e-
studying/network educational process
To make the selection of the methods and means of education
To have the skills of the informational navigation
To plan the perspective stages for the management of the students group (small
group)
To organize using the distance learning technologies of the individually-oriented
approach to the audience
Organizational skills:
To balance the demands of discipline with the students’ needs
To demonstrate to the listeners their personal potentialities concerning to the pro-
vided educational information
To carry out a systematic discussion about the students needs for continuous im-
provement of the distance learning process
To provide the necessary support and assistance to the ENK students;
To organize and conduct network role games
To organize and manage the students’ activity, in a small group, to create the opti-
mal conditions for the development of their independence and competence and
provide pedagogically effective activity of the listeners
To organize the participants’ meetings of thee-studying process
Cognitive skills:
To study physical, psychological and social components of special features of lis-
teners’ individual development, their needs, social self-determination etc.
To analyze individual styles cognitive-educational activity of listeners
To involve modern pedagogical (tutor) experience (learning in cooperation, small
groups method, project method, different-level education, forming evaluation, re-
search, explore methods etc) and creatively apply it in own tutor practice
To master new scientific information in the subject field, methods of teaching and
use rationally in scientific-pedagogical work
To search for facts that stimulate activation of the cognitive activity of students in
informational and educational environment and apply them
To generate new ideas and perspectives of tutor and student activities and apply
modern technologies, forms and methods of distance learning
Communicative skills:
To determine the feasibility of the relationship between subjects of educational
process
286 D. Bodnenko
To coordinate interpersonal relationship between the students in the group or be-
tween students in small groups
To prevent conflict situations that may arise in the process of e-studying, resolve
them
To apply collective activity, cooperate, determine common strategy of activity and
prove its relevance, be able to admit own mistakes
Make simple, easy and tolerant communication with any age, social and ethnic
categories of students.
Perceptive skills:
To understand (by the look of students) incentives of activations of students’ activ-
ity in information-educational environment and be able to apply such knowledge
To be concerned with the inner world of the audience, understand their mental
state. Observe special features of the independent activity of the student in an In-
formation-educational environment during e-studying process.
To improve directly the technology, information saturation, activity, depending on
the needs of the group (the audience)
Suggestive skills:
To master a method of forming a systematic and critical thinking
To form reflection in students as a mean of evaluating their activities with the pur-
pose of further improvement
To affect (emotional and volitional aspect) students with forms, methods and
means of e-studying to create in them a certain mental state, prompting them to
definite actions
Applied skills:
To possess additional hardware, software, psychological educational equipment
To create Web Pages, publications, websites, blogs, wiki, etc.
To have skills to program in specialized environments
Skills in psycho-technical sphere: knowingly and properly use acquisitions from psy-
chology in field of applying network services.
3 Conclusion and Future Work
Psychological, educational requirements for professional competence of university
teachers are based on components of pedagogical skills of teachers of higher educa-
tion institutions and absorb almost all of its functions, duties and skills, but increased
usage of ICT determines the specification of the classification system of psychologi-
cal and educational requirements for information and communication competence of
the university teacher.
According to the results, which we obtained during the research, we can outline
some basic theoretical and empirical achievements of the author in the context of the
study objectives:
Professional competence of teachers of higher education on condition of higher
education tasks’ changes and transition to the society of knowledge were outlined.
System of psycho-pedagogical requirements for information and communication
competence of university lecturer was formed and explored. The system consists of a
The Role of Informatization in the Change of Higher School Tasks … 287
list of skills: informative-communicative, didactic, constitutive, managerial, cogni-
tive, communicative, perceptual, suggestive, applied and the skills of a psychotech-
nique sphere.
Prospects for further scientific research are seen in detailed usage of professional
competence of university teachers in practical activities, including the usage of net-
work services in teaching students of Humanitarian specialties.
References
1. Adrianova, G.: Typologiya Subjectov Distantcionnogo Obucheniya. In: Proc. All-Russian
Remote Teachers Council (2000) http://www.eidos.ru/books/read-room/andrianova1.htm. (in
Russian)
2. Verzhbitsky, V.P.: Distance Education in Russia. http://www.tcde.ru/de/st001.html (in
Russian)
3. Vitvitcka, S.S.: Osnovy Pedagogiky Vyschoi Shkoly. Textbook. Second Edition. Center of
the Education, Kyiv (2006) (in Ukrainian)
4. Goncharenko, S. V.: Ukrainsky Pedagogichny Slovnyk. Second Edition. Lybid, Kyiv
(1997) (in Ukrainian)
5. Bykov, V. Y., Khuharenko, V. N. (eds.): Distance Education Process. Textbook. Mille-
nium, Kyiv (2005) (in Ukrainian)
6. Koycheva, T. I.: Preparation of Future Humanities Teachers as Tutors of Distance Learn-
ing Systems. Manuscript. Konstantin Ushinsky South Ukrainian National Pedagogical
University, Odessa (2004) (in Ukrainian)
7. Kuzminsky, A. I.: Pedagogika Vyshoi Shkoly. Textbook. Knowledge, Kyiv (2005) (in
Ukrainian)
8. Morze, N. V.: Osnovy Iformatciyno-Comynikaciynyh Tehnologiy. Publishing Group,
Kyiv (2006) (in Ukrainian)
9. Oliynyk, V. V.: Distantciyna Osvita za Kordonom ta v Ukraine: Stysliy Analitychniy
Oglyad. CIPPO, Kyiv (2010) (in Ukrainian)
10. Orel, A. M: Tutor s Raznyh Tochek Zreniya. http://www.ou-
link.ru/654/bulletin_654_8/tutor-3.htm (in Russian)
11. Starov, M. I., Chvanova, M. S., Vislobokova M. V.: Psihologo-Pedagogicheskie Problemy pri Dis-
tantcionnom Obuchenii. Distance education, 12(2), 26–30 (2012) (in Russian).
12. Ragan, L. C.: Good Teaching is Good Teaching: an Emerging Set of Guiding Principles
and Practices for the Design and Development of Distance Education. CAUSE/EFFECT
journal, 22(1), http://eee.educuse.edu/ir/library/html/cem9915.html (1999)
13. McKenzie, B., Mims, N., Bennett, E., Waugh, M.: Needs, Concerns and Practices of Online Instruc-
tors. Online Journal of Distance Learning Administration, 3(3) (2011)
1.5 ICTERI Tutorials
UML Profile for MARTE: Time Model and
CCSL
Frédéric Mallet1
Université Nice Sophia Antipolis, Aoste team INRIA/I3S, Sophia Antipolis, France
Frederic.Mallet@unice.fr
Abstract. This 90 minutes tutorial gives a basic introduction to the
UML Profile for MARTE (Modeling and Analysis of Real-Time and
Embedded systems) adopted by the Object Management Group. After
a brief introduction to the UML profiling mechanism, we give a broad
overview of the MARTE Profile. Then, the tutorial shall focus on the
time model of MARTE and its companion language CCSL (Clock Con-
straint Specification Language).
Keywords. UML Profile, Real-Time, Embedded systems, MARTE, CCSL
Key terms. StandardizationProcess, UbiquitousComputing, Concur-
rentComputation, ModelBasedSoftwareDevelopmentMethodology
1 Audience and focus
The targeted audience is academics or industrials interested in high-level mod-
eling with UML and its application to the analysis of real-time and embedded
systems.
The tutorial does not require any preliminary background as it will give a
high-level and broad description of the UML Profile as well as a closer focus on
its time model. To ensure that a large public can follow the presentation, we
should start by a brief overview of UML light-weight extension mechanism, the
so-called profiling mechanism.
2 Topic
The UML profile for Modeling and Analysis of Real-Time and Embedded sys-
tems, referred to as MARTE [1], has been adopted by the OMG in Novem-
ber 2009 and revised in June 2011. It extends the Unified Modeling Language
(UML) [2] with concepts required to model embedded systems.
This ninety minutes tutorial gives a basic introduction to the UML Profile
for MARTE. After a broad view of the Profile, the tutorial shall focus on the
time model of MARTE and its companion language CCSL (Clock Constraint
Specification Language).
290 F. Mallet
2.1 General Introduction to MARTE
Figure 1 shows the general structure of MARTE.
The General Component Modeling (GCM) and Repetitive Structure Mod-
eling (RSM) packages offer a support to capture the application functionality.
GCM defines basic concepts such as data flow ports, components and connec-
tors. RSM provides concepts for expressing repetitive structures and regular
interconnections. It is essential for the expression of parallelism, in both ap-
plication modeling and execution platform modeling; and for the allocation of
applications onto execution platforms.
Fig. 1. Structure of MARTE specification
The Hardware Resource Modeling (HRM) package, which specializes the con-
cepts of GCM into hardware devices such as processor, memory or buses, allows
the modeling of the execution platforms in MARTE. The Allocation (Alloc)
package allows the modeling of the space-time allocation of application func-
tionality on an execution platform. Both the HRM and Alloc packages can be
used with the RSM package for a compact modeling of repetitive hardware (e.g.,
grids of processing elements) and data and computation distributions of a par-
allel application onto such a repetitive hardware.
UML Profile for MARTE: Time Model and CCSL 291
The models described with the previous packages can be refined with tempo-
ral properties specified within the Time package [3]. Such properties are typically
clock constraints denoting some activation rate constraints about considered
components. The concepts of the Time package are often used with the Clock
Constraint Specification Language (CCSL) [4, 5], which was initially introduced
as a non-normative annex of MARTE.
2.2 MARTE Time Model
In MARTE, a clock c is a totally ordered set of instants, Ic . In the following, i
and j are instants.
S A time structure is a set of clocks C and a set of relations on
instants I = c∈C Ic . ccsl considers two kinds of relations: causal and temporal
ones. The basic causal relation is causality/dependency, a binary relation on I:
4⊂ I × I. i 4 j means i causes j or j depends on i. 4 is a pre-order on I, i.e.,
it is reflexive and transitive. The basic temporal relations are precedence (≺),
coincidence (≡), and exclusion (#), three binary relations on I. For any pair of
instants (i, j) ∈ I × I in a time structure, i ≺ j means that the only acceptable
execution traces are those where i occurs strictly before j (i precedes j). ≺ is
transitive and asymmetric (reflexive and antisymmetric). i ≡ j imposes instants
i and j to be coincident, i.e., they must occur at the same execution step, both of
them or none of them. ≡ is an equivalence relation, i.e., it is reflexive, symmetric
and transitive. i # j forbids the coincidence of the two instants, i.e., they cannot
occur at the same execution step. # is irreflexive and symmetric. A consistency
rule is enforced between causal and temporal relations. i 4 j can be refined
either as i ≺ j or i ≡ j, but j can never precede i.
In this paper, we consider discrete sets of instants only, so that the instants
of a clock can be indexed by natural numbers. For a clock c ∈ C, and for any
k ∈ N? , c[k] denotes the k th instant of c.
Specifying a full time structure using only instant relations is not realistic
since clocks are usually infinite sets of instants. Thus, an enumerative spec-
ification of instant relations is forbidden. The Clock Constraint Specification
Language (ccsl) defines a set of time patterns between clocks that apply to
infinitely many instant relations [4].
The uml Profile for marte proposes several specific stereotypes in the Time
chapter to capture ccsl specifications. Figure 2 briefly describes the three main
stereotypes. Boxes with the annotation metaclass denote the uml concepts
on which our profile relies, so-called metaclasses. Boxes with stereotype are the
concepts introduced by marte, i.e., the stereotypes. Arrows with a filled head
represent extensions, whereas normal arrows indicate properties of the intro-
duced stereotypes. Clock extends uml Events to spot those events that can be
used as time bases to express temporal or logical properties. ClockConstraint
extends uml Constraints to make an explicit reference to the constrained clocks.
TimedProcessing extends Action to make explicit their start and finishing
events. When those events are clocks, then a ClockConstraint can constrain
the underlying action to start or finish its execution as defined in a ccsl speci-
fication.
292 F. Mallet
Fig. 2. Excerpt of the MARTE Time Profile
2.3 The Clock Constraint Specification Language
On top of marte clocks, the Clock Constraint Specification Language defines
a set of operators (relations and expressions) [4]. As an example, consider the
clock relation precedence (denoted ≺ ), a transitive asymmetric binary relation
on C: ≺ ⊂ C × C. If lef t and right are two clocks, lef t ≺ right, read ‘lef t
precedes right’, specifies that the k th instant of clock lef t precedes the k th
instant of clock right, for all k. More formally: For a pair of clocks (lef t, right) ∈
C ×C, lef t ≺ right means ∀k ∈ N? , lef t[k] ≺ right[k]. Similarly, let us consider
the transitive and reflexive binary relation on C called isSubclockOf and denoted
⊂ . lef t ⊂ right (read lef t is a sub clock of right) means that for all k, the
instant lef t[k] of lef t coincides with exactly one instant of right. More formally:
lef t ⊂ right means ∀k ∈ N? , ∃n ∈ N? |lef t[k] ≡ right[n]. The relation ⊂ is
order-preserving. All the coincidence-based relations are based on isSubclockOf.
When both lef t ⊂ right and right ⊂ lef t then we say that lef t and right
are synchronous: lef t = right.
A ccsl specification consists of clock declarations and conjunctions of clock
relations between clock expressions. A clock expression defines a set of new clocks
from existing ones, most expressions deterministically define one single clock. An
UML Profile for MARTE: Time Model and CCSL 293
example of clock expression is delay (denoted $: C × N? → C). c $ n specifies
that a new clock is created and is the exact image of c delayed for n instants:
o = c $ n defines a clock o ∈ C such that ∀k ∈ N? , o[k] ≡ c[k + n].
By combining primitive relations and expressions, we derive a very useful
clock relation that denotes a bounded precedence. lef t ≺n right is equivalent
to the conjunction of lef t ≺ right and right ≺ (lef t $ n). The special case,
when n is equal to 1 is called alternation and is denoted lef t ∼ right (reads
lef t alternates with right).
3 Conclusion
The uml Profile for marte is dedicated to the modeling and analysis of real-
time and embedded systems. Its time model relies on a notion of clock borrowed
from the synchronous languages [6] and their polychronous extensions [7]. Those
clocks can be logical or physical. The time model also provides a support to build
causal and temporal constraints to force the clocks to tick according to predefined
patterns. Thus, the evolution of the clocks imposes an execution semantics on
the underlying uml marte model. Whereas the marte time model provides the
notions of clocks and constraints, its companion language, the Clock Constraint
Specification Language provides a syntax to define the constraints themselves.
This brief tutorial introduces the main concepts of marte time model and gives
an overview of ccsl.
Acknowledgments
This work has been partially funded by the PRESTO Project (ARTEMIS-2010-
1-269362).
References
1. OMG: UML Profile for MARTE, v1.1. Object Management Group. (June 2011)
formal/2011-06-02.
2. OMG: UML Superstructure, v2.4.1. Object Management Group. (August 2011)
formal/2011-08-06.
3. André, C., Mallet, F., de Simone, R.: Modeling time(s). In: 10th Int. Conf. on
Model Driven Engineering Languages and Systems (MODELS ’07). Number 4735
in LNCS, Nashville, TN, USA, ACM-IEEE, Springer (September 2007) 559–573
4. André, C.: Syntax and semantics of the Clock Constraint Specification Language
(CCSL). Research Report 6925, INRIA (May 2009)
5. Mallet, F., André, C., de Simone, R.: CCSL: specifying clock constraints with
UML/Marte. Innovations in Systems and Software Engineering 4(3) (2008) 309–
314
6. Benveniste, A., Caspi, P., Edwards, S.A., Halbwachs, N., Le Guernic, P., de Simone,
R.: The synchronous languages 12 years later. Proc. of the IEEE 91(1) (2003) 64–83
294 F. Mallet
7. Le Guernic, P., Talpin, J.P., Le Lann, J.C.: Polychrony for system design. Journal
of Circuits, Systems, and Computers 12(3) (2003) 261–304
8. Mallet, F.: Logical Time in Model Driven Engineering. Habilitation à diriger des
recherches, Université Nice Sophia-Antipolis (November 2010)
Biography
Frédéric Mallet is an associate professor at Université Nice Sophia Antipolis. He
is a permanent member of the Aoste team-project, a joint team between INRIA
and I3S laboratory. He received a PhD in Computer Science in 2000 and his
habilitation degree [8] in 2010. Since 2007, he has been heavily involved in the
definition, finalization and revision of the UML Profile for MARTE1 and has
been a voting member of the successive finalization and revision task forces for
MARTE at the OMG. His main research interests include the definition of models
for the specification of functional and non-functional properties of real-time and
embedded systems. He also develops tools and techniques for the validation and
verification of such models.
1
http://www.omgmarte.org
Ontology Alignment and Applications in 90 Minutes
Vadim Ermolayev1 and Maxim Davidovsky1
1
Department of IT, Zaporozhye National University
Zhukovskogo st. 66, 69063 Zaporozhye, Ukraine
vadim@ermolayev.com, m.davidovsky@gmail.com
Abstract. In this paper, we describe the structure and outline the content of a
short tutorial on Ontology Alignment. The tutorial is planned in three parts
within an overall timeframe of 90 minutes. Part 1 covers the fundamentals of
ontology alignment and offers basic definitions, problem statements and prob-
lem classification based on the span, dynamics, direction, and distribution set-
tings. This material is illustrated by: (i) using a walkthrough example of two
elementary ontologies in Bibliographics domain; and (ii) offering a deeper dis-
cussion of one of the exemplar problems of ontology alignment – ontology in-
stance migration – which has a practical utility for real world applications. The
second part presents a software solution for ontology instance migration prob-
lem. The solution is demonstrated on the pair of Bibliographic ontologies of our
walkthrough example. Part 3 puts ontology alignment in the context of several
categories of applications which are important for the industries and the knowl-
edge economy as a whole. The applications of ontology alignment in those
categories are overviewed and requirements to the solutions are extracted.
Keywords. Ontology, ontology alignment, knowledge-based application, agent,
argumentation, negotiation, information flow, ontology instance migration
Key terms. KnowledgeRepresentation, KnowledgeManagementMethodology,
KnowledgeManagementProcess, KnowledgeTechnology, ICTTool
1 Introduction
This paper outlines the tutorial on the basics and problems of Ontology Alignment.
The material is illustrated by our agent-based solution for ontology instance migration
problem – one of practically important sub-problems in ontology alignment. The de-
mand for applications of ontology alignment in real world applications is also pre-
sented. The tutorial, though given for the first time, is based in parts on our previous
tutorial on Agent-Based Ontology Alignment [1]. This tutorial differs from [1] in the
following: (i) it is broader in scope as covers not only agent-based approaches to align
ontologies; and (ii) it is more oriented to reviewing industrial applications of ontology
alignment and analyzing their requirements to the technology.
296 V. Ermolayev and M. Davidovsky
1.1 Structure and Timeframe
Part 1 targets a broad audience of those who are interested in the problems of ontol-
ogy alignment in general and starts at a relatively basic level. It begins with informal
definition of ontology alignment and puts the problem into the context of the other
knowledge harmonization and integration problems. It further explains the motivation
to study the methods of ontology alignment. Further the basic formalisms for ontol-
ogy alignment are introduced and explained using an incremental approach. The ge-
neric ontology alignment problem is stated first and illustrated by the walkthrough
example. This generic problem statement is further refined by offering a classification
of the types of ontology alignment problems. A particular attention is paid to the on-
tology instance migration problem as a sub-problem of ontology alignment. The time
frame for the first part of the tutorial is 30 minutes1. A standard configuration of pres-
entation equipment is required: 1 beamer, 1 presentation screen, 1 microphone for the
presenter, 1 additional microphone for the questions from the audience.
Part 2 offers a more practical material as it is focused on the presentation of the
agent-based software solution for the ontology instance migration problem. The mate-
rial of this part covers the presentation of the: (i) solution architecture; (ii) methodol-
ogy shaping out the workflow; (iii) software demonstration that migrates instances
from one to the other ontology of our walkthrough example. The time frame for the
second part of the tutorial is also 30 minutes. Part 2 uses two independent presentation
channels: one for the tutor and the other for software demonstration. Therefore it re-
quires an enhanced configuration of presentation equipment: 2 beamers, 2 presenta-
tion screens, 2 microphones for the presenters, 1 additional microphone for the ques-
tions from the audience.
Part 3 is focused on the discussion of the importance of ontology alignment tech-
nology for real world applications. It starts with revisiting the motives to have this
technology in place and proves the necessity of having the solutions for several cate-
gories of ICT applications, particularly in information and knowledge processing. In
fact a review of applications, their specific requirements, and available solutions is
given in this concluding part of the tutorial to provide a holistic, cross-domain view
on the role of ontology alignment as a fundamental technology for today’s knowledge
economy. Similarly to parts 1 and 2, the time frame for part 3 of the tutorial is 30
minutes. Similarly to part 1, part 3 requires a standard set of presentation equipment.
The whole tutorial is therefore given in 90 minutes. A small break could be
planned after Part 2 if the audience wishes to do so for having some discussions or
posing in-depth questions. Though questions are allowed to be posed at any time, all
three parts are planned with 5-minute question and answer sessions at their ends.
1
Timings are given approximately. Small deviations could occur depending on the number of
questions coming from the audience.
Ontology Alignment and Applications in 90 Minutes 297
1.2 A Walkthrough Problem and Example
An example problem of ontology alignment that is used throughout the tutorial for
detailed discussions is the ontology instance migration problem. The problem state-
ment for ontology instance migration is presented in Section 2. The approach and
software for solving this problem is demonstrated in Section 3. The applications that
require the migration of ontology instances are mentioned among those discussed in
Section 4.
Besides that, a very simple and artificial example of two different Biblio ontolo-
gies is used for illustrations throughout the tutorial. The structural schemas and asser-
tional parts of these ontologies are provided in the support material at http://isrg.kit.
znu.edu.ua/a-boa/index.php/A-BOA_Walkthrough_Problem_and_Example.
1.3 Support Materials, Discussions, and Contributions
For additional support materials a reader is advised to visit the A-BOA Wiki
(isrg.kit.znu.edu.ua/A-BOA/) which has been developed for our previous tutorial on
Agent-Based Ontology Alignment [1]. A-BOA Wiki is a Semantic MediaWiki based
collaborative platform and a resource providing teaching content and discussion func-
tionality.
1.4 Motivation to Study Ontology Alignment
The world around us is multi-faceted and polysemic in a sense that a model of the
world developed in the mind of an individual or by a social group may be different
from the model of the others. Knowledge-based systems reflect this fact in their
knowledge representations. However, we do many things across several facets or even
across subject domains. So, the knowledge representations of the corresponding facets
of knowledge representation have to be brought into a harmonized or aligned state to
enable proper communication, coordination or information processing.
Biblio ontologies give a simple example of such different facets, or knowledge
representations, for the same body of knowledge about conference papers. Imagine
that Biblio-2 is the knowledge representation of a conference management system,
while Biblio-1 is the model for a paper repository used by a publisher for book
production. The descriptions of the papers that have been accepted for a conference
have to appear in the publisher’s paper repository. Similarly, the publisher’s informa-
tion about the page limits has to be given to the conference management system to
instruct the authors at proper time. Knowledge representations of Biblio-1 and
Biblio-2 have therefore to be aligned for enabling seamless transformation and
transfer of individual records between these two distributed knowledge-based sys-
tems. The tutorial will teach how such alignments could be done and what the com-
plications in that activity are.
An attendee will learn that an alignment is essentially a result of applying a set of
formal transformations to a knowledge representation – to its structure and individual
assertions. An alignment allows interpreting knowledge that is external to the inter-
298 V. Ermolayev and M. Davidovsky
preter in the same way it interprets its own knowledge schema and assertions. For
example, if an alignment of Biblio-2 to Biblio-1 exists, the publisher, who is
the owner of Biblio-1 may seamlessly import the assertions about the accepted
papers to its production repository. Similarly, an alignment of Biblio-1 to Bib-
lio-2 is required by conference organizers to get the publisher’s information about
publication constraints like page limits.
In a summary, ontology alignment has to be a technology at hand for all those who
develop distributed constellations of knowledge-based systems that require collabora-
tion across the nodes. Building ontology alignments efficiently and effectively is also
important for the management and maintenance of such systems. Indeed, the fact that
you have developed a perfect ontology alignment for your system does not yet allow
you to retire. World changes and these changes are reflected in some facets of knowl-
edge representations sporadically and without informing the other nodes. Hence the
alignment activity has to be repeated in order to bring the whole system to a harmo-
nized state.
2 Basics and Problems of Ontology Alignment
This section of the tutorial presents the formal problem statement and classification of
ontology alignment problems, discusses one of the problem statements – for the on-
tology instance migration problem in more detail.
Following Euzenat and Shvaiko [2], an ontology is formally denoted as a tuple
O C , P, I , T , V , , ,, where C is the set of concepts (or classes); P is the set of
properties (object and datatype properties); I is the set of individuals(or instances);
T is the set of datatypes; V is the set of values; is a reflexive, anti-symmetric and
transitive relation on C C P P T T called specialization, that form partial or-
ders on C and P called concept hierarchy and property hierarchy respectively; is
an irreflexive and symmetric relation on C C P P T T called exclusion; is a
relation over I C V P called instantiation; is a relation over I P I V called
assignment; (the sets C , P, I , T , V are pairwise disjoint). It is also assumed (c.f. [3]),
that an ontology O comprises its schema S and the assertional part A (see also
Fig. 2):
O S , A ; S C , P, T ; A I , V (1)
Ontology schema is also referred to as a terminological component (TBox). It
contains the statements describing the concepts of O, the properties of those concepts,
and the axioms over the schema constituents. The set of individuals, also referred to
as an assertional component (ABox), is the set of the ground statements about the
individuals and their attribution to the schema – i.e. where these individuals belong.
Ontology matching is denoted as a process of discovering the correspondences (or
mappings) between the elements of different ontologies. A mapping (or a mapping
rule [2]) is a tuple m e, e, , n , where: e, e are the elements of C , R, I , T , V of the
Ontology Alignment and Applications in 90 Minutes 299
respective ontologies O and O ; , , , , is a set of relations; and n is a
confidence value (typically in the range of 0,1 ).
Finally, ontology alignment is denoted as the result of applying the discovered set
of mapping rules to the respective ontologies. A generic ontology matching process
and ontology alignment are described and pictured in more detail at http://isrg.kit.znu.
edu.ua/a-boa/index.php/Basic_Definitions_and_Generic_Problem_Statement.
Based on the features of participating ontologies and the span of e, e across
C , P, I , T , V -s of O and O a classification of the problems of finding ontology align-
ments could be outlined and formally stated. Graphical interpretation of some of these
problems is described in more detail at http://isrg.kit.znu.edu.ua/a-boa/index.php/
Classification_of_Ontology_Alignment_Problems. The dimensions along which the
problems are classified are:
Complete (C), structural (S), or assertional (A) alignment
Static (S) versus dynamic (D) aligned ontologies
Bi-directional (B) versus uni-directional (U) alignment
Fully distributed (D) settings versus the presence of a central (C) referee ontology
A generic ontology alignment process may therefore be classified as a complete
static bi-directional alignment using central referee ontology (CSBC). Our walk-
through problem of ontology instance migration could be classified as assertional,
static, uni-directional, distributed (ASUD) ontology alignment problem.
Yet another important feature for classifying ontology alignment processes is the
presence of iterations for the refinement of alignments. All the processes discussed
above are one-shot. However, the resulting alignments may appear to be of insuffi-
cient quality after their evaluation. Iterative ontology alignment processes aim at im-
proving this shortcoming by incorporating the evaluation step and the refinement
cycle in the process – please refer to (http://isrg.kit.znu.edu.ua/a-boa/index.php/
Classification_of_Ontology_Alignment_Problems) for a graphical illustration. Itera-
tive ontology instance migration process is discussed in more detail below. Our
agent-based software prototype toolset for solving this problem is presented in Sec-
tion 3.
One of the practically important ontology alignment problems, especially in fully
distributed and dynamic settings, is the problem of transferring the individuals of one
(source) ontology to the empty assertional part of the other (target) ontology [4].
Let us consider two arbitrary ontologies O s ( S s , As ) and O t ( S t , At ) conceptu-
alizing the semantics of the same universe of discourse U – for example O s and
O t are the two ontologies describing the same subject domain. U could be regarded as
a collection of ground facts: U { f } . Essentially, O s and O t are the interpretations of
U. These ontologies would be considered identical if and only if:
f U int s ( f ) int t ( f ) , (2)
I I
where int I ( f ) is the interpretation of the fact f by the individuals from I of ontology
O.
300 V. Ermolayev and M. Davidovsky
Consequently, an abstract metric of interpretation difference idiff (U , O s , O t ) could
be introduced. The value of idiff will be equal to zero for identical ontologies and will
increase monotonically to one with the increase of the number of f U such
that (int I s ( f ) int I t ( f )) . Hence, idiff 1 iff f U(int I s ( f ) int I t ( f )) . ( 1 idiff )
may further be interpreted [4] as balanced F-measure.
Ontologies O s and O t are structurally different if their schemas differ: S s S t . This
structural difference may be presented as a transformation : S s S t . Transformation
T may be sought in the form of the set of nested transformation rules over the con-
stituents of S s resulting in the corresponding constituents of S t .
Let us assume now that, given two structurally different ontologies O s and O t , the
ABox of O s contains individuals ( I s ), while the ABox of O t is empty ( I t ).
The problem of minimizing idiff (U , O s , O t ) by: (i) taking the individuals from I s ; (ii)
transforming them correspondingly to the structural difference between O s and O t us-
ing T; and (iii) adding them to I t – is denoted as ontology instance migration prob-
lem.
Theoretically ontology instance migration problem can be solved in one shot. In
practice however each of the sub-tasks (ii-iii) may result in the loss of assertions [4].
Therefore an iterative refinement of the solution could yield results with a lower re-
sulting idiff value. Hence, the problem has to be solved using an iterative ontology
alignment process. Essentially, an iterative solution of ontology instance migration
problem develops a sequence of O s states Osts i in a way to minimize the
idiff (U , O s , O t ) in a way that:
idiff (U , Osts i , Ot ) idiff (U , Osts j , Ot ) i j , (3)
where: Osts i is O s in the state after accomplishing iteration i; i, j are iteration numbers.
3 A Solution for Ontology Instance Migration Problem
This section demonstrates our agent-based solution for the ASUD ontology alignment
problem stated above as ontology instance migration problem. This problem has
been chosen as it possesses significant practical interest in real world applications, in
particular for Ontology Engineering and Management in distributed and dynamic
settings [4]. Instance migration in our solution is performed iteratively, so the align-
ment is refined from iteration to iteration.
Many influential publications, for example [5], envision that intelligent software
components, like agents, need to be used together with ontologies for making seman-
tic technologies accepted and effective in open and decentralized scenarios. For such
agent based solutions, comprising industrial applications, the heterogeneity problem is
the challenge that has to be faced. Ontology alignments are a means to solve the chal-
Ontology Alignment and Applications in 90 Minutes 301
lenge. From the other hand agents, being the recipients of ontology alignment solu-
tions, may help solving ontology alignment problems.
For a graphical illustration and more details of a simplified agent-based architec-
ture for solving a generic ontology alignment problem please refer to http://isrg.
kit.znu.edu.ua/a-boa/index.php/Theoretical_Foundations_and_Demonstration. The
architecture introduces the wrapper agents W and W for ontologies O and O respec-
tively. Agent R wraps the central referee ontology O r and helps W and W finding the
proper mappings using O r (a matchmaker function). W and W produce their own
sets of mappings M and M in collaboration with each other (a fully distributed prob-
lem setting) or also in collaboration with R (the problem setting with a central referee
ontology). At the Apply Mappings step M and M are autonomously applied by
W and W to O and O . A problem in developing such an agent-based solution is how
do the agents collaborate and develop these mappings.
The presented solution is based on automated meaning negotiations between agents
[6] as a way to discover structural differences between the schemas of O and O .
Similarly to [7], this approach aims at aligning ontologies by parts (contexts) that are
relevant to a particular negotiation encounter. Negotiations imply iterative monotonic
reduction of semantic distances between the contexts. An agent uses propositional
substitutions which may reduce the distance and support them with argumentation.
The process is stopped when the distance reaches a commonly accepted threshold or
the involved parties exhaust their propositions and arguments. As opposed to the Ar-
gumentation Framework based approaches, this approach addresses the entire process
of semantic reconciliation between ontologies and does not require off-the-shelf map-
pings.
The methodology used in our solution comprises several steps in the workflow.
Steps (I) and (II) correspond to Discover Mappings, step (III) is for Applying Map-
pings, step (IV) corresponds to the step of evaluation and making decision about un-
dertaking one more iteration. Iteration loop however does not involve mappings dis-
covery in our solution. Instead, the mappings are revised manually by a knowledge
engineer based on the list of migration failures in the migration log. Step (V), though
important in practice, is not demonstrated.
Biblio-1 and Biblio-2 are used as examples of O and O . The demonstrated
agent-based solution is evaluated by comparing to our former work [4] where Ontol-
ogy Difference Visualizer (ODV) tool [8] was used for discovering the structural
difference between aligned ontologies.
Ontology instance migration process starts with the step (I) of discovering the
structural difference between O and O . Only TBoxes of the ontologies are used as the
sources. Structural difference is discovered by the SDiff Discovery Engine (SDDE)
[9] – a system of collaborative software agents negotiating on semantic contexts [10]
for finding mappings M : S S . For demonstration purposes discovered structural
difference is visualized using UML extension [8]. The mappings are further written
down by SDDE as instance transformation rules [4] at the subsequent step (II). In-
stance Migration Engine (IME) is invoked at step (III) to perform the instance trans-
formations according to these transformation rules. All the cases in which IME fails
302 V. Ermolayev and M. Davidovsky
to perform the transformation are recorded to the instance migration log. Step (IV)
involves a knowledge engineer who checks the migration log and decides if a refine-
ment is required. If so, he starts the new iteration by refining the set of the transforma-
tion rules based on his analysis of the failure cases and using the rule editor of IMS at
step (II). The refined set of rules is fed to the IMS at step (III). The loop continues
until the knowledge engineer decides that further refinement is not possible, or all the
instances of I s are migrated to I t .
4 Applications of Ontology Alignment
In this part of the tutorial a few selected categories of applications that require align-
ing information or knowledge representations are analyzed. A broader spectrum of
applications is surveyed in [11]. In particular, attention is paid to the requirements
related to ontology alignments that are posed by the applications in each category. A
particular ontology alignment problem fitting to these requirements is also outlined.
A good survey of ontology-based applications is [12]. Ontology matching and
alignment applications are discussed in [2]. Another comprehensive summary of
ontology matching techniques and applications is [13]. In addition to these surveys,
the publications surveying or reporting ontology alignment approaches are for exam-
ple Chuttur [14], Vázquez-Naya et al. [15], Zhdanova et al. [16], Euzenat et al. [17].
Based on these inputs the following several typical application categories are ana-
lyzed in the tutorial with a focus on real world applications.
4.1 Distributed Information Retrieval
Distributed Information Retrieval (DIR) is an important category of applications that
assist retrieving and fusing information from heterogeneous, distributed, and inde-
pendent information resources. Ontologies in DIR are used for representing the struc-
tures of information at different nodes and for translating or transforming user queries
and system responses. In particular, ontologies in DIR are important for extracting
information or knowledge satisfying the semantics and the context of a user query.
Ontology alignments are required:
At query transformation step – for correlating query structure and semantics with
different information resource schemas and metadata and building respective par-
tial queries
At query result fusion step – transforming and putting together the retrieved infor-
mation instances
Hence, a solution of an SSUD ontology alignment problem is required for query
transformation and of an ASUD problem for results fusion and delivery to a user. A
critical requirement at the latter step is high recall as it is important not to miss any
potentially relevant information while irrelevant individuals can be filtered out using
other techniques. One more important requirement to an ontology alignment solution
in DIR is its scalability in terms of the complexity and number of aligned ontologies.
Ontology Alignment and Applications in 90 Minutes 303
4.2 Human-Machine Dialogues
Ontology alignments are used in human-machine interaction for providing mutual
understanding between a user and a processing node. A software agent may represent
a processing node in such interactions as an intelligent wrapper. Ontologies and their
alignments can be used to obtain a formalizable set of requirements, structures, que-
ries, etc. from informal or poorly structured user descriptions. As a rule such dialogs
are run in iterative way. Hence, iterative ontology alignment methods fit to this
category of applications better.
Brasoveanu et al. [18] argue the importance of using generic multimodal ontolo-
gies on the Semantic Web and propose an approach to enhance human-agent interac-
tion based on multimodal ontologies. Guzzoni et al. [19] propose a toolkit-based ap-
proach for modeling human-agent interaction. Their toolset provides a means to
model different aspects of an intelligent assistant such as: ontology-based knowledge
structures; service-based primitive actions; composite processes and procedures; natu-
ral language and dialog structures. Tijerino et al. [20] report a framework for human-
agent collaboration for the purpose of problem solving on the Semantic Web. In hu-
man-machine dialogue scenarios the most critical requirements are adaptability,
integrativity, and scalability that allow enhancing human-machine mutual under-
standing.
4.3 Ontology Evolution, Versioning, Refinement
Ontology evolution, versioning, and refinement are important problems in Ontology
Engineering (OE) and Management (KM) applications. Solutions are required for
adequately representing knowledge in changing domains. Ontology alignment is one
of the enabling technologies in these applications. Indeed, all three problems cope
with transforming a source ontology revision to a target state (revision) that fits to the
requirements causing the transition. Important aspects of this transition are that the
target revision has to: (i) be consistent; (ii) re-use the source as much as allowed by
the requirement of being consistent
Ontology alignments are used both to ensure consistency and maximal possible
degree of re-use. Provided that the source revision is consistent, for proving that the
resulting ontology revision is consistent it is sufficient to build the complete static bi-
directional alignment (CSBC or CSBD problems). For the proper re-use of the source
revision the solution of a uni-directional alignment problem will fit. For example a
typical sub-task in an ontology refinement process is ontology instance migration
from the source revision to the target revision [4]. A balanced combination of appro-
priately high recall and precision is an essential requirement for the instance migra-
tion solution.
4.4 Service Composition
The automation of web service composition or orchestration at run time is a challeng-
ing problem in Service Science which is intensively researched in the last decade. The
304 V. Ermolayev and M. Davidovsky
complexity of the problem is caused by the inherent distributed character of software
systems based on the use of services (for example Web services), the openness of
these systems, and the dynamic character of their configurations and constellations. A
sub-stream of research in the field develops the frameworks for services that inten-
sively use ontologies as service descriptions – Semantic Web Services. Two promi-
nent examples of these frameworks are OWL-S [21] and WSMO/L/X [22] which
however do not fully solve runtime service composition problem. More advanced
approaches exploit collaborative agents as service wrappers for managing services
and service brokers or mediators for manipulating their descriptions in a runtime
composition process (for example [23]). Like in Ontology Engineering and Manage-
ment, a balanced combination of appropriately high recall and precision is an essen-
tial requirement for service composition. The scalability of the solution is also impor-
tant.
The aspects of ontology reconciliation with respect to Web services and their com-
position are elaborated in [24, 25, 26]. An important requirement for such systems is
the capability of adaptation and integration for providing compliant access and mak-
ing the use of aggregate and atomic services more convenient.
5 Learning Outcomes
By the end of the tutorial the participants will:
Learn the basics of ontology alignment that will enable them to understand the
notions of an ontology, ontology mapping, the process of ontology matching, and
the alignment as a result of matching process
Learn the generic ontology alignment problem and the classification of its flavors
based on the features of distributedness, the span of alignment, the direction of
alignment, and the dynamic character of the source ontologies. Specifically, learn
about the ontology instance migration problem as one of the ontology alignment
problems.
Be able to differentiate between one-shot and iterative ontology alignment methods
and judge about the appropriateness of using this or that kind of a method in a par-
ticular setting
Learn about one of the agent-based solutions for ontology alignment (ontology
instance migration problem)
Learn that ontology alignment is a very important, enabling technology for several
kinds of the applications of distributed knowledge-based systems. In particular,
learn which of the requirements of these applications make ontology alignment a
challenging task.
References
1. Ermolayev, V., Davidovsky, M.: Agent-Based Ontology Alignment: Basics, Applications,
Theoretical Foundations, and Demonstration. Tutorial Paper. In: Dan Burdescu, D., Aker-
kar, R., Badica, C. (eds.) Proc. WIMS 2012, 11-22, ACM (2012)
Ontology Alignment and Applications in 90 Minutes 305
2. Euzenat J., Shvaiko P.: Ontology Matching. Berlin Heidelberg, Springer-Verlag (2007)
3. Nardi, D., Brachman, R. J.: An Introduction to Description Logics. In: Baader, F.,
Calvanese, D., McGuinness, D. L., Nardi, D., Patel-Schneider, P. F. (eds.) The Description
Logic Handbook. Cambridge University Press New York, NY, USA (2007)
4. Davidovsky, M., Ermolayev, V., Tolok, V.: Instance Migration between Ontologies having
Structural Differences. Int. J. on Art. Int. Tools. 20(6), 1127-1156 (2011)
5. Berners-Lee, T., Hendler, J., Lassila, O.: The Semantic Web. Scientific American, 284,
28–37 (2001)
6. Ermolayev V., Keberle, N., Matzke, W.-E., Vladimirov, V.: A Strategy for Automated
Meaning Negotiation in Distributed Information Retrieval. In: Y. Gil et al. (Eds.): ISWC
2005 Proc. 4th Int. Semantic Web Conference (ISWC'05), 6–10 November, Galway, Ire-
land. LNCS 3729, pp. 201–215 (2005)
7. Atencia, M., Schorlemmer, M.: Formalising Interaction-Situated Semantic Alignment: The
communication product. In: Proc. of the Tenth International Symposium on Artificial In-
telligence and Mathematics (ISAIM'08), Fort Lauderdale, Florida, USA (2008)
8. Ermolayev, V., Copylov, A., Keberle, N., Jentzsch, E., Matzke, W.-E.: Using Contexts in
Ontology Structural Change Analysis. In: Ermolayev, V., Gomez-Perez, J.-M., Haase, P.,
Warren, P., (eds.) Proc. CIAO 2010, CEUR-WS, vol. 626 (2010)
9. Davidovsky, M., Ermolayev, V., Tolok, V.: Agent-Based Implementation for the Discov-
ery of Structural Difference in OWL DL Ontologies. In: Mayr, H. C., Ginige, A., Liddle,
S. (eds.) Proc. 4th Int. United Information Systems Conference (UNISCON 2012), LNBIP
137, Springer-Verlag, Berlin Heidelberg (2013)
10. Ermolayev, V., Ruiz, C., Tilly, M., Jentzsch, E., Gomez-Perez, G. M., Matzke, W.-E.: A
Context Model for Knowledge Workers. In: Ermolayev, V., Gomez-Perez, J.-M., Haase,
P., Warren, P. (eds.) Proc. CIAO 2010, CEUR-WS, vol. 626 (2010)
11. Davidovsky, M., Ermolayev, V., Tolok, V.: A Survey on Agent-Based Ontology Align-
ment. In: Proc. 4th Int. Conf. on Agents and Artificial Intelligence ICAART’12, pp. 355–
361 (2012)
12. Gargantilla, J., Gomez-Perez, A.: OntoWeb: A Survey on Ontology-Based Applications.
Deliverable 1.6. OntoWeb Consortium IST Project IST-2000-29243 (2004)
13. Scharffe, F., Euzenat, J., Le Duc, C., Mocan, A., Shvaiko, P.: Analysis of Knowledge
Transformation and Merging Techniques and Implementations. KWEB/2004/D2.2.7/0.8
(2007)
14. Chuttur, M. Y.: Challenges Faced by Ontology Matching Techniques: Case Study of the
OAEI Datasets. J. of Information Technology, 3(1), 33–42 (2011).
15. Vázquez-Naya, J. M., Romero, M. M., Loureiro, J. P., Sierra, A. P.: Ontology Alignment
Overview. Encyclopedia of Artificial Intelligence 2009, pp. 1283–1289 (2009)
16. Zhdanova A,.V., de Bruijn, J., Zimmermann, K., Scharffe, F.: Ontology Alignment Solu-
tion. Deliverable D14 v2.0 (2004)
17. Euzenat, J., Laera, L., Tamma, V., Viollet, A.: Negotiation and Argumentation Techniques
among Agents Complying to Different Ontologies. Deliverable D2.3.7, KWEB, v1.0
(2006)
18. Brasoveanu, A., Manolescu, A., Spânu, M. N.: Generic Multimodal Ontologies for Hu-
man-Agent Interaction. Int. J. of Computers, Communications & Control, 5(5), 625–633
(2010)
19. Guzzoni, D., Baur, C., Cheyer, A.: Modeling Human-Agent Interaction with Active On-
tologies. Artificial Intelligence, SS-07-04, 52–59 (2007)
20. Tijerino, Y.A., Al-Muhammed, M., Embley, D.W.: Toward a Flexible Human-Agent Col-
laboration Framework with Mediating Domain Ontologies for the Semantic Web. In: Proc.
306 V. Ermolayev and M. Davidovsky
ISWC 2004 Workshop on Meaning Coordination and Negotiation, Hiroshima, Japan, pp.
131–142 (2004)
21. Martin, D., Paolucci, M., McIlraith, S., Burstein, M., McDermott, D., McGuinness, D.,
Parsia, B., Payne, T., Sabou, M., Solanki, M., Srinivasan, N., Sycara, K.: Bringing Seman-
tics to Web Services: The OWL-S Approach. In: Cardoso, J., Sheth, A. (eds.) Proc.
SWSWPC 2004, LNCS 3387, pp. 26–42 (2004)
22. Roman, D., Keller, U., Lausen, H., de Bruijn, J., Lara, R., Stollberg, M., Polleres, A.,
Feiera, C., Bussler, C., Fensel, D.: Web Service Modeling Ontology. Applied Ontology,
1(1), 77–106 (2005)
23. Ermolayev, V., Keberle, N., Kononenko, O., Plaksin, S., Terziyan, V.: Towards a Frame-
work for Agent-Enabled Semantic Web Service Composition. Int. J. of Web Services Re-
search, 1(3), 63–87 (2004)
24. Li, L., Yang, Y.: Agent Negotiation Based Ontology Refinement Process and Mechanisms
for Service Applications. Service Oriented Computing and Applications, 2, 15–25 (2008)
25. Paurobally, S., Tamma, V., Wooldridge, M.: A Framework for Web Service Negotiation.
ACM Transactions on Autonomous and Adaptive Systems (TAAS), 2(4) (2007)
26. Huang, J., Zavala, R., Mendoza, B., Huhns, M. N.: Reconciling Agent Ontologies for Web
Service Applications. In: Proc. of Multiagent System Technologies: Third German Con-
ference (MATES-05). LNAI 3550, pp. 106–117 (2005)
Biographies
Vadim Ermolayev is an associate professor at the Department of Information Tech-
nologies (IT) of Zaporozhye National University and the lead of Intelligent Systems
Research Group. He is also a research consultant in Semantic Technologies, Intelli-
gent Software Systems, Distributed Artificial Intelligence. The research projects he
took part in were focused on: intelligent systems and knowledge representations for
enterprises; business and informal process dynamics; intelligent distributed informa-
tion retrieval; the confluence of agent-based systems and Semantic Web services;
ontology engineering, evolution, and refinement; performance management in engi-
neering design. Alignment of knowledge representations was one of important topics
in those projects.
Maxim Davidovsky is a PhD candidate at the Department of Mathematical Modeling
(MM) of Zaporozhye National University. He also works for the Laboratory of Web-
based Technologies and Distance Learning and is the member of Intelligent Systems
Research Group at the Department of IT. Maxim received his MSc degree in applied
mathematics and accomplished his postgraduate course in mathematical modeling and
computational methods at the Department of MM. His research interests are in dis-
tributed and decentralized knowledge-based systems and software development. The
focus of his current research activity is agent-based ontology alignment and instance
migration specifically in distributed and decentralized settings.
Part 2. ICTERI Workshops
2.1 2nd International Workshop
on Information Technologies
in Economic Research
(ITER 2013)
Foreword
We would like to offer you the section including a ten papers we have selected for the
2-nd instance of our International Workshop on Information technologies in economic
research (ITER 2013) which has been organized as a session in the technical the 9-th
International Conference on ICT in Education, Research, and Industrial Applications:
Integration, Harmonization, and Knowledge Transfer (ICTERI 2012) held at Kherson,
Ukraine on June 19-22, 2013.
The necessity to support decisions by the means of IT at different levels of busi-
ness process any organization, to verify economical hypotheses and usage of gained
knowledge in the learning process requires the use of IT to process relevant informa-
tion.
Skills of analytical information processing for decision making can effectively be
realized only by using information and communication technologies.
The large numbers of economic studies is not supported by modern mathematical
framework and ICT, leading to poor quality of the research on both the micro-,
macro-and industry levels.
Creation of ITER is intended to familiarize researchers with modern ICT and
mathematical methods of information processing in areas such as
Business process management: business process for firms, suppliers, customers,
information systems in small and medium business, IT-innovations in management
process, R&D company, business intelligence approach, management in virtual or-
ganization, e-commerce, cloud technology in business, e-governance.
Quantitative methods in economics: econometrics research on micro- and macro
level, business process economical modeling, software package for economic re-
search, modeling industry mergers and acquisitions, finance modeling.
IT education for economists: business informatics curricula, IT-professional
training for economical organizations, software programs for economic education,
mobile technologies in economical education, e-learning for economists, business
games.
Under international level research many scientist and researchers use the Ukrainian
and Russian language, which greatly limits the possibilities for the world community
to become familiar with published papers. This restricted the number of accepted
papers till 10 of the 20 provided for ITER:
Business process management
Decision Supporting Procedure for Strategic Planning: DEA Implementation for
Regional Economy Efficiency Estimation
Applying of Fuzzy Logic Modeling for the Assessment of ERP Projects Efficiency
Matrix Analogues of the Diffie-Hellman Protocol
Binary Quasi Equidistant and Reflected Codes in Mixed Numeration Systems
Quantitative methods in economics
Mechanism Design for Foreign Producers of Unique Homogeneity Product
Are securities secure: Study of the Influence of the International Debt Securities on
the Economic Growth
310
How to make high-tech industry highly developed? Effective model of national
R&D investment policy
IT education for economists
Econometric Analysis on the Site “Lesson Pulse”
Mathematical Model of Banking Firm as Tool for Analysis, Management and
Learning
Features of National Welfare Innovative Potential Parametric Indication’ Informa-
tion-Analytical Tools System in the Globalization Trends’ Context
Only timely and qualitative preparation of an economic study will provide recom-
mendations and suggestions for decision-makers, to promote the efficient use of mate-
rial and budgetary resources in organization.
June, 2013 Tanya Payentko
Sergey Kryukov
Vitaliy Kobets
Binary Quasi Equidistant and Reflected Codes in Mixed
Numeration Systems
Evgeny Beletsky1 and Anatoly Beletsky1
1
Department of Electronics, National Aviation University of Kiev,
1, av. Cosmonaut Komarov, 03680, Kiev, Ukraine
ebeletskiy@gmail.com, abelnau@ukr.net
Abstract. The problem of constructing quasi equidistant and reflected binary
Gray code sequences and code in a mixed factorial, Fibonacci and binomial
numeration systems is considered in the article. Some combinatorial construc-
tions and machine algorithms synthesis sequences, based on the method of di-
rected enumeration are offered. For selected parameters of sequences all quasi
equidistant (for individual cases - reflected) codes with Hamming distance
equal to 1 are found.
Keywords. Reflected codes, quasi equidistant sequence, Hamming distance
Key terms. Research, CodingTheory, MathematicalModelling
1 Introduction
Coding theory is one of the most important areas of modern applied mathematics.
Beginning of the formation of mathematical coding theory dates back to 1948, when
it was published a famous article by Claude Shannon [1]. The growth of codes origi-
nally was stimulated by tasks of communication. Later constructed codes found many
other applications. Now codes are using to protect data in a computer memory, cryp-
tography, data compression, etc.
The work is devoted to a rather small, but extremely important for applications
subset of so-called quasi-equidistant and reflected codes. The class of quasi equidis-
tant codes are sequences of uniform (i.e., containing the same number of bits) of bi-
nary code combinations in which any adjacent (neighboring) code sets (words) are at
the same Hamming d distance equal to a fixed number of natural numbers (i.e. d =
1, 2, …) [2]. Equidistant sets include such codes in which any two words (code com-
binations) are at the same distance d [3].
Finally, we shall refer to the reflected subset quasi equidistant codes with distance
d =1, the formation of which is based on the principle of mirror reflection? [4]. But if
we restrict ourselves to only one mirror, the code sequence will contain the original
sequence, after which is the same sequence just re-written in reverse order, which is
312 E. Beletsky and A. Beletsky
unacceptable, since it leads to code repetition. The elimination of repetition can be
provided by initial expansion of the number of digits combinations. The essence of
the "mirror" reflection of the expansion is explained below as an example of canoni-
cal reflected Gray codes and in other sections of this article.
The main objective of this study is to develop algorithms for constructing quasi-
equidistant and reflected binary Gray codes as well as code sequences in a mixed
factorial, Fibonacci and binomial bases. The method of direct enumeration is the base
of algorithms of computer sequences synthesis.
2 Basic of Number System
The history of discrete mathematics and computer science is directly related to the
development and introduction of newer principles of representation and encoding
digital information, which are based on the numeration system of numbers. By a nu-
meration system we understand the way of image sets of numbers using a limited set
of characters that form its alphabet, in which the characters (elements of the alphabet)
are located in the established order, occupying a certain positions [5]. Any numeration
system should be composed of a finite set of non-negative numbers — a range that it
encodes. It always includes the number 0 and then follows the natural numbers start-
ing with 1 [6].
There are various numeration system (as well as methods for their classification),
whose number is constantly growing. All systems can be divided into the following
main classes: positional, not positional and mixed. In the positional numeration sys-
tems the same numeric characters (digit) has different meanings in its description
depending on the location (level) where it is resides.
By positional numeration system is generally understood the p numeration system,
which is defined by an integer p 1 — is called a base of numeration system. Un-
signed integer N in p numeration system is represented as a finite linear combina-
tion of powers of
n
N k p k , (1)
k 1
where k are integers satisfying the inequality 0 k ( p 1) , n the number of
digits of the number. The simplest examples of positioning systems (1) can be binary,
decimal, and other numeration systems.
In no positional numeration systems the value which indicated by the digit does
not depend on the position in a number. At the same time the system may impose
restrictions on the position of numbers, for example, that they are in descending order.
The Roman and many other systems belong to not positional systems.
The mixed numeration system is a generalization of the p system, and often refers
to the positional numeration systems. The base of mixed numeration system is an
increasing sequence of numbers pk , k 1, 2, , and each N number is presented
like linear combination:
Binary Quasi Equidistant and Reflected Codes … 313
n
N k pk ,
k 1
there are some restrictions exist for k coefficient.
One of the known examples of the mixed system is a factorial numeration system,
in which the bases are the sequence of factorials pk k! . Another commonly used
Fibonacci numeration system is a system that is based on Fibonacci numbers. The
Binomial system in the form in which it is presented in the relevant section of this
article, we will also include to a mixed numeration system.
A positive integer is depicted in an arbitrary numeration system as a sequence of
symbols N n n 1 k 21, where N - the number representation in this
numeration system, besides each k symbol takes rk bit in general case (if binary
alphabet is using).
Note the following general characteristics of quasi equidistant codes with Ham-
ming distance d 1 . Let’s agree each code sequence starts with zero code. And as
result of this agreement the following code after the zero code should be placed with
weights 1 and 2, and Further weight codes must alternate even (E) — odd (O) under
the scheme
012OEOE E(O) . (2)
Scheme (2) is a symbolic form of the tree sequence code combinations. Let’s ne
and no to be the amount of even and odd code words in a sequence. If the sequence
(2) ends up with odd code combination this means ne no , and if even —
ne no 1 . This becomes evident:
Statement 1. Inequality
0 ( ne no ) 1 , (3)
is a necessary (but not always sufficient) condition for the construction of quasi equi-
distant codes.
3 Sequences of Gray Codes
Classic Gray codes [7] may be called canonical, since for arbitrary length sequence of
combinations are not only quasi equidistant, but also reflected. Let’s G ( n ) se-
quence of n-bites classical Gray codes. To construct ( n 1) bites reflected Gray
Codes, let’s us note as Grc ( n 1) codes, it is just enough to prefix for each source
code G ( n ) the 0 digit and 1 to the left of code group G R (n ) constructed by reflected
(reflex or reverse) mirror of G ( n ) sequence, i.e.
314 E. Beletsky and A. Beletsky
Grc ( n 1) 0G ( n ) ||1G R ( n ) , (4)
where || - is a symbol of concatenation (conjunction of sequences).
According to (4), Grc ( n 1) G ( n 1) and as a result sequences of Gray codes
of G ( n ) number of digits n 2 are both quasi equidistant and reflected, and besides
the line of reflection goes through 2n 1 and (2n 1 1) code combinations. On the
basis of the canonical code G ( n ) , n 2 , the equidistant Gray codes can be con-
structed. For example, Tab. 1 show the three 12-bit code quasi equidistant sequences,
one of which corresponds to the canonical version of the Gray code.
The first six variants of sequences in the table constructed of canonical option 1 as
a result of a variety column rearrangement saving the Hamming distance d 1 of
related code combinations. Variants 7-12 are formed as a result of inverse none zero
rearrangements of code combinations from appropriate variants 1-6.
Table 1. Three bit quasi equidistant Gray code
Variants of sequence
1 2 3 4 5 6 7 8 9 10 11 12
000 000 000 000 000 000 000 000 000 000 000 000
001 100 100 001 010 010 100 001 010 010 001 100
011 110 101 101 110 011 101 101 110 011 011 110
010 010 001 100 100 001 111 111 111 111 111 111
110 011 011 110 101 101 110 011 011 110 101 101
111 111 111 111 111 111 010 010 001 100 100 001
101 101 110 011 011 110 011 110 101 101 110 011
100 001 010 010 001 100 001 100 100 001 010 010
The first six variants of sequences in the table constructed of canonical option 1 as
a result of a variety column rearrangement saving the Hamming distance d 1 of
related code combinations. Variants 7-12 are formed as a result of inverse none zero
rearrangements of code combinations from appropriate variants 1-6. As follows from
Tab. 1 the only variants 1 (canonical) and 6 of Gray codes belong to a set of three
bites reflected codes. At the same time each three bite sequence by (4) statement pro-
duce subset of four bite reflected Gray codes. Thereby it is true:
Statement 2. All amounts L(отG ) ( n ) of reflected Gray codes of n number of digits is
defined by
n, if n 2;
L(rcG ) (n 1)
2n !, if n 3.
Binary Quasi Equidistant and Reflected Codes … 315
3 Factorial Sequence
The integer positive number N in factorial number of numeration system can be
represented as
n
N k k !, 0 k k (5)
k 1
where k 1, 2,, n; 0 k k . Extended form of (5) statement is
N n n ! n 1 (n 1)! 2 2! 1 1! , (6)
Statement (6) is so called numerical, or digital, function [8] of factorial system.
There are first 120 decimal numbers (Tab. 2) defined by their k coefficients in facto-
rial numeration system.
Table 2. Binary representations of decimal numbers of factorial numeration system
N N k Fakt N N k Fakt N N k Fakt N N k Fakt N N k Fakt
0 0 24 100000 48 1000000 72 1100000 96 10000000
1 1 25 100001 49 1000001 73 1100001 97 10000001
2 10 26 100010 50 1000010 74 1100010 98 10000010
3 11 27 100011 51 1000011 75 1100011 99 10000011
4 100 28 100100 52 1000100 76 1100100 100 10000100
5 101 29 100101 53 1000101 77 1100101 101 10000101
6 1000 30 101000 54 1001000 78 1101000 102 10001000
7 1001 31 101001 55 1001001 79 1101001 103 10001001
8 1010 32 101010 56 1001010 80 1101010 104 10001010
9 1011 33 101011 57 1001011 81 1101011 105 10001011
10 1100 34 101100 58 1001100 82 1101100 106 10001100
11 1101 35 101101 59 1001101 83 1101101 107 10001101
12 10000 36 110000 60 1010000 84 1110000 108 10010000
13 10001 37 110001 61 1010001 85 1110001 109 10010001
14 10010 38 110010 62 1010010 86 1110010 110 10010010
15 10011 39 110011 63 1010011 87 1110011 111 10010011
16 10100 40 110100 64 1010100 88 1110100 112 10010100
17 10101 41 110101 65 1010101 89 1110101 113 10010101
18 11000 42 111000 66 1011000 90 1111000 114 10011000
19 11001 43 111001 67 1011001 91 1111001 115 10011001
20 11010 44 11110 68 1011010 92 1111010 116 10011010
21 11011 45 111011 69 1011011 93 1111011 117 10011011
22 11100 46 111100 70 1011100 94 1111100 118 10011100
23 11101 47 111101 71 1011101 95 1111101 119 10011101
316 E. Beletsky and A. Beletsky
Let’s mark Ф( k ) sequence of n bite factorial codes. In the case where number of
digits of code combination from code set Ф( k ) less than k , it is prefixed with required
amount of zeros. Let’s Фd ( k ) sequence of quasi equidistant k bite factorial codes
with Hamming distances among related combinations equal to d . Based on data from
Tab. 2 it is easy to create (Tab. 3) sequences Ф1 ( k ) for k 1 (singular case), and also
k 2 and k 3 created by columns rearrangement of base sequences (variant 1).
Table 3. Sequences of quasi equidistant Factorial Codes
Ф1 ( k )
k 1 k2 k 3
1 1 2 1 2 3 4 5 6
0 00 00 000 000 000 000 000 000
1 01 10 010 010 100 100 001 001
11 11 011 110 101 110 011 101
10 01 001 100 001 010 010 100
101 101 011 011 110 110
100 001 010 001 100 010
Table 3 illustrates one possible method of synthesis of quasi equidistant codes. Its
idea is in the following. At the very first stage the base sequence of quasi equidistant
codes of n number of digits is created by means of some method (for example, the
method of direct search which is examined below). On the second stage a variety of
all possible rearrangements of base sequence columns (check out Tab. 3, the corre-
spondent values are of number 1) is done which results in formation of n! different
quasi equidistant codes. And finally on the third stage the sequences which contain
restricted code combinations are excluded from n! sequences. Such combinations are
110 codes from Tab. 3 highlighted with bold type. So from six three bite sequences
the only two generate quasi equidistant factorial sequences. Starting from k 4 apart
from quasi equidistant sets it is possible to create reflected factorial codes Фrc ( k ) .
Starting from k 4 apart from quasi equidistant sets it is possible to create reflected
factorial codes Фrc ( k ) . The algorithm of reflected codes creation depends on their
number of digits. In particular, here is easily provable by direct verification.
Statement 3. The set of uniform reflected factorial codes defined by recurrence re-
lation
Фrc ( k ) 0Ф1 ( k 1) ||1Ф1R ( k 1) ,
Let’s discuss the problem of synthesis of quasi equidistant factorial codes with a
number of digits n 4, 7 . So taking the data from Tab. 3 let’s construct a preliminary
weights distribution of n bite code combinations resulting in Tab. 4. The amount of
codes with even and odd weights in current table for all variants n are satisfying
inequality (3) and this means, that all required conditions for quasi equidistant facto-
rial codes creation are met.
Binary Quasi Equidistant and Reflected Codes … 317
Schema (2) of uniform codes Ф(4) weights interchanges, according to Tab. 4, is
012O2O2O2O2O (7)
Table 4. Weights distribution of code words Ф ( n )
Code The bit of code combinations
weight n4 n5 n6 n7
0 1 1 1 1
1 4 5 6 7
2 5 9 14 20
3 2 7 16 30
4 2 9 25
5 2 11
6 2
ne 6 12 24 48
no 6 12 24 48
In all 12 24 48 96
At that from 5 odd elements (O) of sequence (7) two elements are equal 3 and the
rest – 1. Which means, that there are ten possible variants of quasi equidistant facto-
rial code trees of number of digits n 4 , from whose the one, for depiction, is shown
on Fig. 1.
Fig. 1. Tree Ф1 (4) of sequence 012321232121
The symbolic form of the tree of code combination sequence Ф1 (5) can be repre-
sented by schema
012OEOEOEOEOEOEOEOEOEOEO , (8)
One of variants is shown on Fig. 2.
Fig. 2. The variant of tree sequence Ф1 (5)
318 E. Beletsky and A. Beletsky
Let’s go to validation to the whole amount of trees variants Ф1 (5) . First of all pay
attention (Fig. 2) the code combinations with weight of 4 must reside between codes
with weights equal 3. This is required to provide a distance between related combina-
tions equal to 1. Merge code pairs with weights equal to 3 among whose code with
weights equal to 4 are reside. By that we can get rid of two code pairs with weights 3
and 4 in column n 5 Tab. 4 and schema (8) rewrite as
012O2O2O2O2O2O2O2O2O . (9)
There are group of nine odd (O) code combinations which contains four codes with
weight equal to 1 and five with weight equal to 3 in the schema (9). It is evident the
126 variant of not complete trees of sequence Ф1 (5) exists, equal to number of nine
by four combinations. And now take into consideration that in each of 126 variants of
symbolic form (9) because of the operation, inversed to “merge” operation described
above, it is possible to restore entire schemas of trees (8). Because of 10 possible
methods of inverse operation means the entire amount of trees Ф1 (5) construction
equal to 1260. Performing by the same method validation of amount of trees LФ (6)
of Ф1 (6) sequences we get LФ (6) =1513512. With increasing of number of digits n
the complexity of combinatorial validation LФ ( n ) and amount of trees Ф1 ( n ) dra-
matically increases. For example, all 10 variants of trees Ф1 (4) are shown in Tab. 5.
Ф1 (4)
Table 5. Trees
№ Tree variant № Tree variant
1 012323212121 6 012123212321
2 012321232121 7 012123212123
3 012321212321 8 012121232321
4 012321212123 9 012121232123
5 012123232121 10 012121212323
First of all we construct ranged by weights v sequence of uniform codes
Ф (4) (Tab. 6).
Table 6. Ranged Ф (4) codes
Code weight v
№
0 1 2 3
1 0000 0001 0011 0111
2 0010 0101 1011
3 0100 0110
4 1000 1001
5 1010
Binary Quasi Equidistant and Reflected Codes … 319
In correspondence with a schema of sixth tree variant (Tab. 5) the first two code
sequences, which will be called layers of tree branch, choose 0000 and 0001 codes.
We could choose 0010 layer instead of 0001. The third layer to choose would be a
code with weight equal to 2, the one which consist of 0001 code with Hamming dis-
tance equal to 1. Suitable ones are codes in columns with 1, 2 and 4 numbers of Tab.
6. The code with smaller number will be considered as a base, the rest – alternative.
Keep moving the same way with codes choosing for Ф1 (4) sequence, using the
schema of chosen tree, we have a Tab. 7.
Table 7. Synthesis of of Ф1(4) branch
№ Code weight Base code Alternative code
1 0 0000
2 1 0001 0010
3 2 0011 0101 1001
4 1 0010
5 2 0110
The ninth layer of tree under synthesis should be a code with weight equal to 2,
moreover it must reside from previous code with distance equal to 1.
But there is no such a code, which were not used in Tab. 6. In order to cope with
this deadlock we will do the following. We will go up through columns of and will do
a substitution in this row with a nearest alternative code located from the right of it. In
this case we should substitute base code 0011 with alternative code 0101 and after-
wards continue the synthesis procedure for Ф1 (4) . An example of quasi equidistant
codes Ф1 (4) synthesized by method of direct enumeration is shown in Tab. 8.
Table 8. Ф1 (4) Sequences, correspondent to 012321212321 tree
Number The branch of the tree
Tree
of tiers 1 2 3 4 5 6 7 8 9 10
0 0 0000 0000 0000 0000 0000 0000 0000 0000 0000 0000
1 1 0001 0001 0010 0010 0010 0010 0100 0100 0100 0100
2 2 1001 1001 0011 0011 1010 1010 0101 0101 1100 1100
3 3 1011 1101 1011 1011 1011 1011 1101 1101 1101 1101
4 2 0011 0101 1010 1010 0011 0011 1100 1100 0101 0101
5 1 0010 0100 1000 1000 0001 0001 1000 1000 0001 0001
Number The branch of the tree
Tree
of tiers 1 2 3 4 5 6 7 8 9 10
6 2 1010 1100 1001 1100 0101 1001 1001 1010 0011 1001
7 1 1000 1000 0001 0100 0100 1000 0001 0010 0010 1000
8 2 1100 1010 0101 0101 1100 1100 0011 0011 1010 1010
9 3 1101 1011 1101 1101 1101 1101 1011 1011 1011 1011
10 2 0101 0011 1100 1001 1001 0101 1010 1001 1001 0011
11 1 0100 0010 0100 0001 1000 0100 0010 0001 1000 0010
320 E. Beletsky and A. Beletsky
4 Fibonacci Sequences
Fibonacci codes are generalized concept of classical binary code [9]. Any nonnega-
tive integer N 0, 1, 2, … can be exclusively represented by a numerical Fibo-
nacci function
N n Fn n 1Fn 1 k Fk 2 F2 1F1 . (10)
Besides the sequence { k } in (1) doesn’t contain pairs of neighbor unities which
are provided by equivalent conversion called “folding’ operation: 011 100 . This
operation makes it possible to represent Fibonacci number as so called “minimal”
form, the code combination of which will have minimal weight.
For example, [10],
01111011001 10011100001 10100100001 . (11)
The codes which are underlined in example (11) are codes for which folding opera-
tion was performed. As it follows from this example the folding operations resulted in
weights decreasing of code combinations. Namely, the amount of units in the final
code is less than in the original one.
Using the folding operation it is easy to come to a representational algorithm of
multidigit binary Fibonacci numbers. As an example let’s consider a method of repre-
sentation of natural sequence of decimal numbers (including zero) by four digits
numbers of Fibonacci codes. We need to agree to label code numbers from right to
left assuming the smaller (the very right) number the correspond to number 1, then
number 2 and so on. We choose such a coding method of first three decimal numbers
0, 1 and 2:
010 0000 ;
110 0001 ; (12)
210 0010.
A conversion from decimal number k10 to ( k 1)10 number in Fibonacci codes (la-
bel them as Fk and Fk 1 correspondingly) will be performed using a rule: if there is
0 in a smaller position Fk then it is substituted with 1 in Fk 1 code. If there is 1 in a
smaller position Fk then this 1 goes to the second position and writes as 0 in a
smaller position. This rule is using in system (12) while conversion from F1 to F2 .
Let’s represent number 310 with Fibonacci code. But before we go, following the
rule described above we will get code 310 00011 which by folding operation would
be represented in its minimal form
Binary Quasi Equidistant and Reflected Codes … 321
310 0100. (13)
According to statements (12) and (13), the smaller positions of Fibonacci codes are
using for decimal numbers 1, 2 and 3 representations correspondingly. Those values
are generalized by the following recurrent block synthesis algorithm of binary Fibo-
nacci sequences. Let’s F ( k ) is a set of Fibonacci numbers of the same length in-
cluding 0. Then we have:
Statement 4. A set of k bite Fibonacci numbers of the same length is defined by
recurrent correlation
F ( k ) 10 || F ( k 2) . (14)
The proving of just formulated statement can be easily performed by a method of
direct verification. In the right part of (14) the F ( k 2) set is consisted of ( k 2)
position numbers.
From this it is followed that if any subset of Fibonacci numbers, included in
F ( k 2) , contain digits the number of digits of whose are less than k 2 then those
numbers are prefixed with required amount of zeros. Algorithm (14) is right for any
value k 2 . Indeed, if k 2 then
F (2) 10 || F (0) .
As long as F (0) set is empty then F (2) set contains the only Fibonacci digit 10,
which corresponds to decimal digit 210 .
There are Fibonacci codes for limited sequence of decimal numbers calculated us-
ing recurrent formula considering initial condition (12) in Tab. 9. Zeros, which are
located to the left of bigger unit in Fibonacci coders, have been removed.
You can see values n in column F of Tab. 9, equal to number of codes which
can be created by a fixed number of binary positions. For example, F 3 means the
four bite combinations, which contain 1 in its older position, can be created three
Fibonacci codes. Writing down the values from F column we will get sequence 1, 1,
2, 3, 5, 8, 13,… which is classical sequence of Fibonacci numbers.
Now go to estimation of variants of quasi equidistant Fibonacci code trees. For this
purpose based on data from Tab. 9 let’s create a preliminary table of distribution of
code combinations weights, included in F ( k ) , k 4, 7 , (Tab. 10). By analysis of
data from Tab. 10 we have the following conclusion. Quasi equidistant sequences of
four digit Fibonacci numbers are end up with code combinations with weight of 1,
five or six number of digits with weight of 2 and seven numbers of digits with odd
weight equal to 1 or 3.
322 E. Beletsky and A. Beletsky
Table 9. Fibonacci numbers
k10 Fk F k10 Fk F k10 Fk F
0 0 13 100000 21 1000000
1 1 1 14 100001 22 1000001
2 10 1 15 100010 23 1000010
3 100 16 100100 24 1000100
2
4 101 17 100101 25 1000101
5 1000 18 101000 26 1001000
6 1001 3 19 101001 8 27 1001001 13
7 1010 20 101010 28 1001010
8 10000 29 1010000
9 10001 30 1010001
10 10010 5 31 1010010
11 10100 32 1010100
12 10101 33 1010101
Table 10. Distribution of code combinations weights F ( k )
All code Number of code digits ( k )
combinations 4 5 6 7
0 1 1 1 1
1 4 5 6 7
2 3 6 10 15
3 1 4 10
4 1
nч 4 7 11 17
nн 4 6 10 17
All together 8 13 21 34
It is not that complicated to perform a calculation LF ( k ) of quantity of variants for
quasi equidistant Fibonacci sequence F1 ( k ) trees.
The result of this calculation for chosen k parameters is shown in Tab. 11.
Table 11. Power of tree subset F1 ( k )
Amount of tree variants F1 ( k )
k 4 5 6 7
LF ( k ) 1 5 126 205920
For reflected Fibonacci codes it is right the following
Statement 5. A set of even k bite reflected Fibonacci codes is defined by recurrent
correlation
Binary Quasi Equidistant and Reflected Codes … 323
Фот ( k ) 00 F1 ( k 2) 10 F1R ( k 2) , (15)
where F1R ( k ) sequence is inversed to F1 ( k ) , i.e. the sequence of quasi equidistant
codes F1 ( k ) written in reverse order.
As an example (Tab. 12) of calculated using a computer a branch of one tree
F1 (6) .
Table 12. Sequences F1 ( k ) of tree 012321232123232121212
Number The branch of the tree
Tree
of tiers 1 2 3 4 5 6 7 8
0 0 000000 000000 000000 000000 000000 000000 000000 000000
1 1 000001 000001 000010 000100 000100 001000 001000 010000
2 2 000101 010001 100010 000101 010100 001010 101000 010001
3 3 010101 010101 101010 010101 010101 101010 101010 010101
4 2 010100 000101 001010 010001 000101 101000 100010 010100
5 1 000100 000100 001000 000001 000001 100000 100000 000100
6 2 100100 100100 101000 100001 100001 100001 100001 000101
7 3 100101 100101 101001 100101 100101 101001 101001 100101
8 2 100001 100001 001001 100100 000000 001001 001001 100100
9 1 100000 100000 000001 100000 100000 000001 000001 100000
10 2 100010 100010 010001 100010 100010 010001 010001 100010
11 3 101010 101010 010101 101010 101010 010101 010101 101010
12 2 101000 101000 000101 101000 101000 000101 000101 101000
13 3 101001 101001 100101 101001 101001 100101 100101 101001
14 2 001001 001001 100001 001001 001001 100100 100100 100001
15 1 001000 001000 100000 001000 001000 000100 000100 000001
16 2 001010 001010 100100 001010 001010 010100 010100 001001
17 1 000010 000010 000100 000010 000010 010000 010000 001000
Number The branch of the tree
Tree
of tiers 1 2 3 4 5 6 7 8
18 2 010010 010010 010100 010010 010010 010010 010010 001010
19 1 010000 010000 010000 010000 010000 000010 000010 000010
20 2 010001 010100 010010 010100 010001 100010 001010 010100
324 E. Beletsky and A. Beletsky
5 Binomial Sequences
There are many known methods for binomial codes creation and based on them –
binomial sequences [11]. We will consider two ways of even binomial codes synthesis
in this unit. First of them we will call an “algorithm A. Borysenko”, and the second
one an “algorithm of A. Beletsky”, which is called as alternative algorithm here in
after.
The whole idea of first algorithm of uneven binary binomial codes, which correlate
to algorithm of full summarized binomial arithmetic, is described in [12], page 124.
Of course any uneven binary code can be converted to even code of n number of dig-
its (length). For this purpose it is just enough to prefix the code combination such
amount of zeros so the common number of digits became equal to n .
To construct algorithms of binomial arithmetic by Borysenko it is enough to define
two parameters k and n, the first one defines the maximal amount of units in codes,
the second one by value r n 1 , defines the maximal length of uneven binomial
number. A decimal zero in Borysenko’s binomial code is written down as l n k of
zeros, the range P of binomial numbers is defined by formula Fmax P 1 . Here are
a number of examples of binomial numbers Bx (algorithm A. Borysenko), creation
whose correspond to decimal value x (Tab. 13).
Table 13. Variants of binomial number sequences
n = 6, k = 4 n = 6, k = 2 n = 6, k = 3
x Bx x Bx x Bx x Bx x Bx x Bx
0 00 10 11010 0 0000 10 10000 0 000 10 1000
1 010 11 11011 1 00010 11 10001 1 0010 11 10010
2 0110 12 11100 2 00011 12 1001 2 00110 12 10011
3 01110 13 11101 3 00100 13 101 3 00111 13 10100
4 01111 14 1111 4 00101 14 11 4 0100 14 10101
5 100 5 0011 5 01010 15 1011
6 1010 6 01000 6 01011 16 11000
7 10110 7 01001 7 01100 17 11001
8 10111 8 0101 8 01101 18 1101
9 1100 9 011 9 0111 19 111
Let’s label B ( n, k ) sequence of binomial numbers created by Borysenko’s algo-
rithm. From analysis of Tab. 4 we get the following conclusion.
Statement 6. Direct and inverse binomial sequences are linked with correlation
R
B ( n, k ) B ( n, n k ) ,
Binary Quasi Equidistant and Reflected Codes … 325
R
where B ( n, n k ) sequence of binomial codes, which first of all is written in re-
R
verse order to codes in B ( n, k ) and secondly each position of B ( n, n k ) forms by
result of inversion (i.e. substitution of 0 to 1 and vice versa) of corresponding posi-
tions B ( n, k ) .
Let’s find out a possibility of quasi equidistant codes B1 ( n, k ) creation based on
set of binomial numbers B ( n, k ) . For this purpose using the data from Tab. 13 lets
create a table of code combination weights distribution (Tab. 14) included in B ( n, k )
set. According to data from Tab. 14 and also values ne and no comparison, received
for many other parameters n and k , we can conclude the inequality (3) for codes
B ( n, k ) is not true and as sequence it is true
Table 14. Distribution of code combination weights B ( n, k )
Weight of code B (6, 4) B (6, 2) B (6, 3)
combination
0 1 1 1
1 2 4 3
2 3 10 6
3 4 10
4 5
nч 9 11 7
nн 6 4 13
All together 15 15 20
Statement 7. Binomial codes do not form quasi equidistant sequences.
Let’s move to creation of alternative binomial codes. Introduce numeric function
B nCnn n 1Cnn11 k Ckk 1C11 (15)
where
k k ( k 1)( k 1 l )
Clk ,
l l!
- binomial coefficient which is equal to number of k and l combinations. The coef-
ficients k are defined by a correlation k 0, k / 2 , in which x means
rounding of number x to the nearest integer above.
Series (15) is presented in form of binary coefficients k for each of who’s the
limited number of positions equal to number of digits and required for binary value
k / 2 representation is assigned.
Coefficient unambiguously defines the value of monomial k Ckk , as it is shown in
Tab. 15 (in which for example purpose the value k 7 is chosen).
326 E. Beletsky and A. Beletsky
Table 15. An example of monomial series calculation (16)
7 0 1 2 3 4
C77 1 7 21 35 35
7C77 0 7 42 105 140
For a sequence of binomial codes created by numerical function (15), let’s intro-
duce a label B ( n, r ) in which n parameter will be called a power of a function, and
r order of function, which is equal to coefficient n . A fragment of binomial
codes is shown in Tab. 16.
Table 16. The sequence of binomial numbers B (4, 2)
N 3 2 1 N 4 3 2 1
0 0 10 1 0 1 1 1
1 1 11 1 1 0 0 1
12 1 1 0 1 0
2 1 0 13 1 1 0 1 1
3 1 1
14 1 0 0 0 1 0
4 1 0 1 15 1 0 0 0 1 1
5 1 1 0 16 1 0 0 1 0 1
6 1 1 1 17 1 0 0 1 1 0
18 1 0 0 1 1 1
7 1 0 0 1 19 1 0 1 0 0 1
N 3 2 1 N 4 3 2 1
8 1 0 1 0 20 1 0 1 0 1 0
9 1 0 1 1 21 1 0 1 0 1 1
In order to decide a question regarding the possibility of quasi equidistant binomial
sequences creation let’s create a table of a set of code combinations weights (Tab.
17).
Table 17. Distribution of weights of code combinations B1 ( n, r )
Amount of digits of binomial sequence
Weight
3 4 5 6 7 8 9 10
0 1 1 1 1 1 1 1 1
Amount of digits of binomial sequence
Weight
3 4 5 6 7 8 9 10
1 2 2 2 2 2 2 2 2
2 3 5 5 6 6 6 6 6
3 1 2 4 9 9 12 12 12
Amount of digits of binomial sequence
Weight
3 4 5 6 7 8 9 10
Binary Quasi Equidistant and Reflected Codes … 327
4 2 4 7 15 15 17
5 2 12 12 23
6 4 8 29
7 2 18
8 4
Even 4 6 8 1 14 26 30 57
Odd 3 4 6 11 13 26 28 55
In all 7 10 14 22 27 52 58 112
Sign + – – + + + – –
As an example check Tab. 18, where results of quasi equidistant codes creation by
a method of direct enumeration based on one of trees for B (4,2) is shown.
Table 18. Results of computer code synthesis B1(4,2)
Number The branch of the tree
Tree
of tiers 1 2 3 4 5 6 7 8
0 0 000000 000000 000000 000000 000000 000000 000000 000000
1 1 000001 000001 000001 000001 000001 000001 000001 000001
2 2 000101 000101 000101 000101 000101 000101 000101 000101
3 3 100101 100101 100101 100101 100101 100101 100101 100101
4 4 100111 100111 100111 100111 100111 100111 100111 100111
5 3 100011 100011 100110 100110 100110 100110 100110 100110
6 2 000011 100010 100010 100010 100010 100010 100010 100010
7 3 001011 101010 100011 100011 101010 101010 101010 101010
8 2 001010 001010 000011 000011 001010 001010 001010 001010
9 3 011010 011010 001011 001011 001011 011010 011010 011010
10 4 011011 011011 011011 101011 011011 011011 011011 011011
11 3 011001 011001 011001 101001 011001 001011 011001 011001
12 2 001001 001001 001001 001001 001001 001001 001001 001001
13 3 101001 101001 101001 011001 101001 101001 101001 101001
14 4 101011 101011 101011 011011 101011 101011 101011 101011
15 3 101010 001011 101010 011010 100011 100011 100011 001011
16 2 100010 000011 001010 001010 000011 000011 000011 000011
17 1 000010 000010 000010 000010 000010 000010 000010 000010
18 2 000110 000110 000110 000110 000110 000110 000110 000110
19 1 000010 000010 000010 000010 000010 000010 000010 000010
Number The branch of the tree
Tree
of tiers 1 2 3 4 5 6 7 8
20 2 000110 000110 000110 000110 000110 000110 000110 000110
21 3 000111 000111 000111 000111 000111 000111 000111 000111
22 4 010111 010111 010111 010111 010111 010111 010111 010111
23 3 010110 010110 010110 010110 010110 010110 010110 010110
328 E. Beletsky and A. Beletsky
A feature of alternative binomial codes is that they do not allow creating quasi
equidistant codes in a full manner as it is visible from Tab. 18. In particular, for all
sequences shown in Tab. 18, the latest codes (highlighted) reside from previous codes
with a Hamming distance equal 3 but not 1, as it is required for sequence B1 (4,2) .
This feature of alternative binomial codes is visible in all possible variants B1 ( n, r ) .
6 Conclusions
The main result of this research is formation of generalized conditions for quasi equi-
distant and reflected codes existence which are produced by even consistent binary
code combinations in a mixed numeration systems. Except of Gray codes the Fibo-
nacci, factorial and binomial codes with Hamming distance between related code
combinations equal to 1, are also included in a set of such codes. The main method
for synthesis of quasi equidistant codes is a method of computer direct enumeration.
The results of this research can be easily generalized and applied for cases where
Hamming distance is more than 1.
References
1. Shannon, C. T.: A Mathematical Theory of Communication. Bell. Syst. Tech. J., 27, 379 –
423, 623 – 656 (1948)
2. Efimenko, V. V., Karpjuk, B. V., Stukalin, Iu. A.: An Algorithm for Synthesis of Binary
Quasi Equidistant Codes. Journal of Acad. Sience, USSR, AVTOMETRIJA, 5, 109–115
(1968) (In Russian)
3. Bogdanov, G. T., Zinovjev, V. A. Todorov, T. J.: On the Construction of Quasi
Equidistant Codes. Journal of Problems of Information Transmission, 43(4), 13–36 (2007)
(In Russian)
4. Beletsky, A. Y., Beletsky E. A.: Quasi Equidistant Codes. NAU Publishing , Kiev (2008)
(In Russian)
5. Banja, E. N., Selivanov, V. L.: About the Features of the Construction of Various Number
Systems. Journal of NTUU "KPI" Informatics, Management and Computer Science, 49,
68–73 (2008) (In Russian)
6. Borysenko, A. A., Cherednychenko, V. B.: Number Systems in Computing. Bulletin of the
SSU, Engineering Series, 4, 162–177 (2009) (In Russian)
7. Grey, F.: Pulse Code Communication, Pat. USA, № 2632058 (1953)
8. Borysenko, A. A.: Discrete Mathematic. Textbook publishing house SSU (2007) (In
Russian)
9. Stahov, A. P.: Codes of Golden Proportion. Radio Communication, Moscow (1984) (In
Russian)
10. Stahov, A., P.: Fibonacci Codes, http://goldenmuseum.com/1010FibCodes_rus.html
11. Zanten, A. Ja.: Binomial System and Enumerations of Combinatorial Objects. Journal of
Discrete Analysis and Operations Research, Series 1. 6, 12–18 (1999) (In Russian)
12. Borysenko, A. A.: Binomial Count. Theory and practice. Publishing house SSU (2004) (In
Russian)
Mechanism Design for Foreign Producers of Unique
Homogeneity Product
Vitaliy Kobets1
1
Kherson State University, 1, 40 rokiv Zhovtnya Street, 73000, Kherson, Ukraine
vkobets@kse.org.ua
Abstract. Paper concerns to impact on custom receipts of duty rate changing
from single to differentiated ones by customs house for foreign producers. To
get maximal custom receipts for achieving of social goal state may introduce
differentiated duty rates for foreign producers of unique product. Success of this
state policy will depend on effectiveness of incentive compatibility conditions
for these producers.
Keywords. Mechanism design, single duty, differentiated duty, custom policy,
social choice function
Key terms. MechanisnDesign, RevelationPrinciple, SocialGoal, Mathemati-
calModel, IncentiveCompatible
1 Introduction
Mechanism is a mathematical structure, modeling an institute and determining the set
of rules, regulating actions accessible to the participants and determining as partici-
pants strategies in given communication system are transformed in results. In the
absence of co-operation mechanism between participants the final result can substan-
tially differ from social optimal one. A mechanism implements given objective func-
tion, realizing it on participants types space [2; 8].
Mechanism structure includes [7]:
1. Social choice function (SCF is a final result demanded by the society)
2. Implementation mechanism (realization of SCF by the payoffs and distributive
functions of product and money);
3. Revelation mechanism of participants types (by a social planner);
4. Motivating mechanism (it is intended to make conditions for revelation of true
information by participants about their types [6]).
Objective function F is a composition of messages and result h (fig.1).
330 V. Kobets
Fig. 3. Mechanism M and objective function F
Mechanism can enforce to cooperation rules, when participants accomplish ac-
tions, violating set rules [4]. Two extremes of mechanisms’ types are centralized
(planned system) and decentralized (such as competition market), between them are
continuum numbers of other mechanisms.
The decentralized mechanism (saving confidentiality mechanism) implies the pri-
vate expenditure (for collection and verification of information reliability) [1]. Exis-
tent mechanisms can be complemented or substituted by the new ones, for example,
by the means of legislation changes.
Reasons of new mechanism introduction:
Revelation of unsatisfactory aspects activity of existent economic systems or insti-
tutes (market failures)
Established economic system gives advantage only for certain participants
Mechanism tasks:
To ground social choice function with the desirable characteristics for society
To develop compatible conditions for participants to reveal their true types (reser-
vation price, costs etc.)
To make implementation process of social choice function by the help of chosen
mechanism (direct or indirect)
The direct mechanisms provide direct transfer of the truthful private information
about their types by the agents to the public planner (not realistic mechanism). The
indirect mechanisms create motive, under which to the agents more profitable to open
true information, than to conceal or to distort it (more realistic mechanism) [10].
During organizing of customs mechanism, as well as any other, to the participants
concerns of social planner (government) and agents (payers of the customs tax). The
agent is selfish person, who has private information only about him or her own type
(for example, personal income, costs, profit).
Economic environment is exogenous variable, given by nature or received from the
last periods (competition type, technology, rule of custom policy). In model neither
the agents, nor mechanism designer do not know prevailing environment. Mechanism
designer knows: (i) class of environments, for which should be developed the mecha-
nism; (ii) desired criterions for SCF [5; 9].
Mechanism Design for Foreign Producers of Unique Homogeneity Product 331
SCF represents criterions for result estimation, but not a means of goal achieve-
ments as mechanism does.
For customs house SCF mapping types space (average costs of production for im-
porters) in results space (custom receipts). The participant type (average costs) de-
fines its message (invoice cost of the goods), which causes final result (custom re-
ceipts).
So the purpose of customs mechanism can be maximization of receipts from cus-
toms taxes in the state budget under creation of the appropriate motivating system for
the importers (increase of the invoice price, preferential duty rates regime).
This paper has a following structure. We make a literature review in this, first, part.
Problem statement and basic assumptions of model are presented in the second part.
Part 3 deals with main results for participants under fixed and differentiated duty
rates. Last part concludes.
2 Problem Statement
Search of effective ways of state budget replenishment by the means of indirect taxes
requires introduction of flexible duty customs for foreign producers foreseen by the
proper government laws in relation to payment of custom payments. Criteria, after
which the state aims to set the duty rate on import commodities, and to foresee protec-
tionism principle for domestic producers, profitableness principle for the state and
utility principle, for domestic consumers.
Peculiarity of optimization for import duty rate is that foreign producers, forming a
competition domestic market, will maximize own profits, taking into account a mar-
ket price [3], whereas for state size of custom rate depends on invoice cost of com-
modity, which can be corrected by a custom house in the direction of increase and
have to corresponds to prevailing (equilibrium) market price.
Product invoice price indicates cost of commodity, which transfers through custom
border of Ukraine. If the invoice price indicated in freight customs declaration below
of average price in the base of Government custom service of Ukraine, there is the
rise of product price to average level before getting customs clearance for product.
From the customs value of product duty and VAT is counted, that countries are trans-
ferred in a budget.
Customs takes place as follows:
B t TRN ,
where B – duty sum; TRN – part of product invoice price, that exceeds an un-
taxable size (in UAH); t – duty rate (in per cent) from the product invoice price,
which now in Ukraine is equal 10%.
For construction of model, that describes co-operation of foreign producers and
customs we assume:
332 V. Kobets
n foreign firms produce homogeneous product, which is supplied to the domestic
market and has no domestic analogues;
between firms there is quantitative Cournot competition;
cost functions of all firms are linear on production quantities (constant scale re-
turn), and reverse domestic market demand function is linear on the quantity of
foreign products;
information about average costs of foreign firms and domestic market demand is
uniformly (symmetrically) distributed between all participants (foreign producers,
domestic consumers and government).
Participants’ objective functions:
Foreign producers:
Total cost of producer i consists of variable cost (fixed cost we assume zero in long-
run period, vi is average (variable) cost of producer i) and duty sum:
TCiF vi qi t P qi , where Pf P is invoice price of unit product.
Profit of producer i: iF P qi vi qi t P qi or
iF (1 t ) P qi vi qi
i q 0
max , i 1,..., n , t is endogenous vari-
able, i.e. duty rate determined by government.
1. State (Ukrainian Income and Duty Ministry)
n
Tax proceeds to state budget is: B t Pf q .
i 1
i
2. Domestic market:
n
Reverse linear function of domestic demand is P b c Q b c q ,
i 1
i
where P is market price of product, b - maximal price of foreign product on do-
mestic market (under zero import supply).
3 Results
3.1 Custom Receipts Model Construction for Fixed Duty Rate
3.1.1 Producer profit maximization
After substitution of market price to profit function of producer i we obtain:
n
iF (1 t ) b c qi qi vi qi , i 1,..., n . First order condition (FOC)
i 1
Mechanism Design for Foreign Producers of Unique Homogeneity Product 333
iF
of profit give us: (1 t ) b 2c qi c qi vi 0 . Similarly we
qi j i
get partial derivatives for profit function of all producers. Algebraic transformation
yields:
v1
2 q1 q2 ... qn b c (1 t ) ,
v2
q1 2 q2 ... qn b ,
c (1 t )
...................................................,
vn
q1 q2 ... 2 qn b c (1 t ) .
Solving of system by matrix approach give optimal value of foreign producer sales
(duty rate 0 t 1 ):
1 (n 1) v j n v
qj b , j 1,..., n , (1)
(n 1) c 1 t
n
v i
where v i 1 - average product cost of all foreign producers. If in equation
n
(1) average cost of producer j lower than average cost of all producers: v j v , then
after increasing of duty rate t , its optimal sales will rise. And vice versa: if v j v ,
then optimal sale of producer j will decrease.
Total quantity of foreign producers with using of (1) will equal:
n
n (1 t ) b v
Q qj . (2)
j 1 c (n 1) (1 t )
Thus growth of duty rate always will lead to decreasing of total quantity of unique
dQ
good for foreign producers at domestic market: 0.
dt
3.1.2 Budget Custom Receipts Maximization
The receipts from foreign producers’ duty customs after substitution of total sales of
import quantity in expression (2) will give:
t 0
B t Pf Q max , where Pf const – product unit invoice price.
334 V. Kobets
First order condition for maximization of custom receipts is determined by condi-
dB
tion 0 or equivalent to following equation: bt 2 2b t b v 0 , from
dt
here equilibrium duty rate will be equal:
v
t 1 . (3)
b
Equilibrium single duty rate (3) will have inverse relation with average cost of all
foreign producers and direct relation with maximal domestic product price.
Thus the invoice price of product will be set at a level P b c Q or taking into
accoun (2) and (3) we will get the equilibrium indexes of invoice price and sales
accordingly:
Pf*
b b n v , Q n b b v .
*
(n 1) c (n 1)
.
Farther from expression B t Pf Q we will define that the equilibrium
(maximal) custom sum will form:
b v b n v .
2
n b
B* (4)
c (n 1) 2
Consider dependence between equilibrium duty state and custom receipts on fig.2.
Dependence of custom receipt on duty rate
100000
80000
60000
custom receipts
B
40000
Bd
20000
0
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
%
10
15
20
25
30
35
40
45
50
55
60
65
70
75
80
85
90
-20000
duty rate, %
Fig. 2. Laffer curve – dependence between custom receipts and duty rate
Mechanism Design for Foreign Producers of Unique Homogeneity Product 335
(n=10, b=40, v=5,25, c=0,01, P=24,04) t=73%.
3.2 Custom Receipts Model Construction for Differentiated Duty Rate
3.2.1 Producer profit maximization
Profit of foreign producer I is presented by next expression:
(1 ti ) P qi vi qi
i
F qi 0
max , where ti is differentiated duty rate for
foreign producer i. FOC for profit function gives ( i 1,..., n ):
iB
(1 ti ) b 2c qi c qi vi 0. .
qi j i
Similarly we obtain partial derivatives for profit functions of others foreign pro-
ducers.
b v1
2 q1 q2 ... qn c c (1 t ) ,
1
b v1
q1 2 q2 ... qn ,
c c (1 t2 ) .
...................................................,
b v1
q1 q2 ... 2 qn c c (1 t ) .
n
System solving by matrix approach give optimal sales values for foreign producers
on domestic market (duty rate 0 ti 1 , i 1,..., n ):
1 nvj v
qj b i , j 1,..., n . (5)
(n 1) c 1 t j i j 1 ti
3.2.2 Budget custom receipts maximization
Receipts from taxation of differentiated duty rates for foreign producers of homogene-
ity products will equal:
n
Bd Pf ti qi
ti 0,i 1,..., n
max , where Pf const – invoice price per
i 1
unit product for foreign producers.
FOC for maximization of custom receipts is defined by following n conditions:
dBd
0 , where i 1,..., n . Obtained n-equation system with n unknown duty
dti
336 V. Kobets
rates ti after equivalent algebraic transformations define reaction curves (6)
ti fi (t i ) , which demonstrate dependence duty rate of i-th producer ti and duty
rates of all its rivals t i . In this function duty rates for foreign producers have to
change in a same direction. Thus increasing optimal duty rate by one of the producers
requires rising of duty rates for all others foreign producers.
vi (1 t j )
j i
ti 1 , i 1,..., n . (6)
vj
b
j i 1 t j
Such adjustment change of duty rates will proceed until the equilibrium size of
each duty rates will not be set.
System solving of n equation formed from functions (6) gives the following sizes
of equilibrium duty rates for foreign producers:
v1
ti 1 , i 1,..., n . (7)
b
Obtained result shows reverse dependence between the average cost and size of
dti
duty rate for import product: 0 and shows direct dependence between
dvi
maximal price of domestic market and duty rate.
From expression (7) follows that more effective producers (with average cost
lower than industry average cost) will be assessed after the higher duty rate, than less
effective ones for maximization of custom receipts. Now equilibrium invoice price
and quantity sale with using of expression (7) and linear function of domestic demand
will be set at the appropriate levels:
n
n
b b vi b n b vi
Pfd * i 1 , Qd* i 1 .
(n 1) c (n 1)
It is important circumstance that after differentiation of duty rates for the foreign
producers, import of product on domestic market will drop Q Q , that will result
d* *
in rising of price for consumers. Additionally, possibility of charging lower duty rates
for one producer and higher for another ones will generate corruption actions. To
prevention it, necessary objective indexes for differentiation of these rates. Expedi-
ence of differentiated rate introduction will arise only after condition of increase of
custom receipts Bd B in comparison with the fixed duty rate (fig. 3).
Mechanism Design for Foreign Producers of Unique Homogeneity Product 337
82500
82000
81500
Bd
B
81000
80500
80000
71% 71% 72% 72% 73% 73% 74% 74% 75% 75% 76%
Fig. 3. Comparison of influencing of the differentiated and fixed duty rates on customs receipts
(n=10, b=40, v=5,25, c=0,01, P=24,04)
4 Conclusions
So if as to differentiation of tariffs to take public accountant reports from the financial
records audit of firm in part of total cost forming, it will decrease possibilities of re-
alization of unfounded duty rates differentiation by custom house.
At the same time more effective producers will be interested not to disclosure in-
formation about true total cost with purpose to drop size of duty rates. Less effective
producers vice versa will have motives to reveal its total cost, which below than aver-
age cost per unit.
Thus if less effective producers will prove that the effective ones gave false infor-
mation, it will become foundation for the rise of duty rates to more effective produc-
ers and decrease of duty rate for less effective. Thus such custom policy will allow to
the state will put information transaction cost about producers from itself on less ef-
fective producers. From one's part, more effective producers will have motives to
prove that less effective producers set too high the size of its inefficiency.
Collusion between all foreign producers about non-disclosure information about
own costs will be highly unlikely when number of foreign producers will be grow and
collusion will be high-probability when firm concentration will be high.
To receive maximal custom receipts for achieving of social aim state may imple-
ment differentiated duty rates for foreign producers of unique product which depend
from producers’ cost. Success of this state policy will depend on effectiveness of
incentive compatibility conditions for these producers, which mean extracting of true
information about cost from foreign producers by the means of firms cross-sectional
audit.
338 V. Kobets
References
1. Dilip, M.: Decentralization, Hierarchies, and Incentives: A Mechanism Design Perspec-
tive. Journal of Economic Literature, 44, 367–390 (2006)
2. Jehle, G. A., Reny, P. J.: Advanced Microeconomic Theory. Prentice Hall, New York
(2005)
3. Williamson, O. E.: Markets and Hierarchies: Analysis and Antitrust Implications. Free
Press, New York (1975)
4. Maskin, E.: Mechanism Design: How to Achieve Social Goals, HSE, Moscow (2009) (In
Russian)
5. Nikolenko, S. I.: Mechanism Design Theory, Binom, Moscow (2009) (In Russian)
6. Archibald, G. C.: Information, Incentives and the Economics of Control. Cambridge Uni-
versity Press, London (2005)
7. Narahari, Y., Garg, D., Narayanam, R., Prakash H.: Game Theoretic Problems in Network
Economics and Mechanism Design Solutions. Springer Series in Advanced Information
and Knowledge Processing (AIKP). Springer-Verlag London Limited, London (2009)
8. Hurwicz, L., Stanley, R.: Designing Economic Mechanisms. Cambridge University Press,
Cambridge (2006)
9. Izmalkov, S., Sonin, K., Yudkevich, M.: Mechanism Design Theory. Questions of Eco-
nomics, 1, 4–26 (2008) (In Russian)
10. Bergemann, D., Stephen, M.: Robust Mechanism Design. Econometrica, 73, 1771–1813
(2005)
11. Myerson, R.: Game Theory Analysis of Conflict. Harvard University Press, Cambridge
(1997)
Features of National Welfare Innovative Potential
Parametric Indication Information-Analytical Tools
System in the Globalization Trends’ Context
Elena Lazareva
Southern Federal University, 105, str. Bolshaya Sadovaya, 344006, Rostov-on-Don, Russia
el_lazareva@mail.ru
Abstract. In the article innovation-reproductive and rent-generating function of
national welfare is exposed, necessity and real ways of national welfare innova-
tive potential look-ahead analytical estimates methodology and tools’ revision
in a context of globalization trends are offered, complex analysis of the results
of system parametristic indication of the strategy of national welfare develop-
ment in the innovation economic growth interests on the author’s set of instru-
ments ground is conducted.
Keywords. National welfare, innovation as a new form of combining industrial,
intellectual and social resources; innovation rent; information-analytical tools
system
Key terms. NationalWelfare, CorporateModel, SpatialStrategy
1 Introduction
In a context of the modern economic development model the essence of national wel-
fare is expressed in new aspects – it becomes not only the accumulated re-iterative
reproduction process result, but also is converted into the integrated innovation-
oriented economic growth resource-factor. This conversion is connected with world
and national economic systems movement towards innovative «knowledge econ-
omy», competition gravity center transference to the science, education, innovative
activity sphere, non-material actives role in economic reproduction process increas-
ing.
The resource-provided countries have the export-raw model of economy. Their de-
velopment may be characterized in comparison with other countries by the rough,
spasmodic rate, mainly caused by considerable raw materials prices and economic
instability. Such development is inevitably accompanied by the problems which brake
economic modernization and its social and innovative orientation. On the contrary,
the development of the countries which realize the policy of human capital quality,
340 E. Lazareva
national well-being, high technologies increment provides advantages in world socio-
economic evolution, raises competitiveness of national «intellectual» economy.
Increasing human development quality importance for economic growth generating
and competitiveness initiated the mounting interest of economists to the subjective
factor (the human capital) role in production progress. It gradually promoted the na-
tional welfare parameters (at first – the individual, especially economic; later – the
social, public) inclusion into the economic dynamics resource supply research system.
The globalization accompanied by substantial capital mobility and national
economies openness increasing transforms the national welfare economic content and
display forms in reproduction process, modernizes its structure and functions in the
conditions of transition to the innovative-focused economy.
These tendencies find reflection in the new long-term economic trend research
methodology – the methodology which equally considers society and economy inter-
ests. The national welfare becomes the major productive forces element and the inte-
grated institutional condition of the human capital reproduction.
The world financial-economic crisis, showing critical dependence of the national
economies upon mobile global resources (financial, information, technologic re-
sources) and, in particular, exposing the fact that the dynamics of the Russian GDP is
still to a considerable extent determined by the external factors of the conjuncture,
made topical the problem of finding internal, innovative resources-sources for devel-
opment. The present situation requires the internal innovative resources of social-
economic development, first of all, existing resources of the national welfare, the
reserves of which in Russia are still unused in full due to the underdeveloped nature
of the institutions of their conversion into competitive factors of production, active
usage.
In this light, especial topicality is attributed to the issues related to technical-
methodological analysis of the national welfare resources in the system of the global
competitive resources, to determination of their role in the process of social innova-
tive reproduction, conditions and mechanisms of their conversion into the innovative
factors of production as well as integrated evaluation of the human capital of the
country, its efficient usage and higher rate of innovation oriented development of the
economic subjects and of economy overall.
The strategy of the economic subjects policies’ economic-oriented modernization
has to be based, due to the aforementioned facts, upon evolutional-cyclical, informa-
tional-innovative paradigm of the economic development theory and upon resource
analysis, in accordance with which national welfare in the postindustrial society plays
the role of an integrated resource for the innovative economic trends. One witnesses
not only a different nature of the input of national welfare into the reproduction proc-
ess, but its various composition, i.e. apart from traditional material elements, which
have cost measurement (revenue level, volume and structure of the personal con-
sumption fund etc.), greater importance is attributed to its social elements – level and
quality of education of the population, level of its health, housing conditions, degree
of security within the society, quality of the social-ecologic habitat, social capital,
social-economic mentality, condition of general and spiritual culture in the society, set
of the symbolic benefits etc., which do not have market cost and, often, which have
the nature of social benefits, i.e. they create general social conditions for fulfillment
of a person, for creative freedom.
Features of National Welfare Innovative Potential Parametric Indication … 341
In accordance with the above, the economic subjects’ innovative social-economic
policies have to comprise not only the innovative processes direct support strategies
and mechanisms, but also person-oriented, comfortable general social conditions for
innovation-oriented development of economy creation, realized in the form of wel-
fare, better life standards and insuring efficient reproduction of the human capital [1].
Orientation towards bigger human and social capital and, consequently, bigger in-
vestments in the anthropo-social capital in the process of intellectual resources social
reproduction as integral parts of the national welfare, constitute the basis for forming
its innovative, resource-reproducing functions. It follows that the principal problem of
continued innovative economic development consists of the national welfare into the
innovation-initiating human and social resources – factors of production inherent to
the “economy of knowledge” social-economic conversion mechanisms. Such the
national welfare components institutional conversion into the innovative resources,
human and social benefits economic composition is their comparative advantages
(competitiveness) capitalization within the framework of countries’ integration in the
world economic relations,, i.e. transformation of said advantages into the source of
the added value and objects of the global companies, business, integrated structures,
states innovative activities.
The innovative rent, which is received due to national welfare reproduction and its
conversion into the innovative-intellectual production economic resources, constitutes
an economic basis of the innovation oriented development. Within the framework of
the cluster theory, “network economy” – the innovative economic rent plays the role
of the result of the national welfare components, situated in the country, efficient
usage. The variety of the innovative rent categories is due to different categories of
benefits – resources of national welfare, which constitute the source of the rent forma-
tion.
The research is based upon such founders of the innovation-oriented economic de-
velopment theory as D. Bell, A. Buzgalin, V. Inozemtsev, N. Kondratjev, S. Kuznets,
B. Kuzyk, G. Mensch, B. Milner, R. Nizhegorodtsev, D, North, V. Ovchinnikov, J.
Osipov, D. Tis, E. Toffler, J. Schumpeter, J. Jakovets etc.
The national welfare potential evaluating scientific basis of the welfare economy in
the innovation-oriented development of economy system is considered in the works of
the following authors: J. Bentham, S. Valentej, L. Walras, A. Marshall, L. Nesterov,
V. Pareto, A. Pigou, A. Smith, J. Hicks, L. Erhard etc., who analyze the problems of
the benefits value, of wealth formation, its distribution, conditions for market balance
as a principal factor for social welfare, problems of harmonizing individual
and social welfare judging by different criteria. D. Buchanan, J. Galbraith, J. Mill, W.
Eucken, J. Rawls, V. Cherkovets, R. Ehrenberg etc. analyze a great number of social-
economic factors, which affect the social welfare growth in the market economy.
Different aspects of the national welfare role identifying in the of innovation-
oriented economic development system are researched in the works of P. Aguillon, R.
Barro, A. Varshavskij, J. Vinslav, S. Glazjev, I. Diskin, J. Coleman, V. Kostjuk, D.
Lvov, V. Makarov, N. Moisejev, N. Rimashevskaja, D. Rodrik, S. Rosefielde, K.
Salomon, A. Sen, R. Solow, J. Stiglitz, M. Todaro, F. Fukuyama etc. Study of their
works allowed specifying scientific interpretations of national welfare from the new
institutional evolutional cyclical paradigm of economic development point of view.
342 E. Lazareva
The nature and specificity of national welfare systems functioning taking into account
their correlations with innovative development of economy and its different institu-
tional structures were researched by A. Auzan, P. Drucker, V. Ivanter, G. Kleiner, A.
Prokhorovskij, V. Tambovtsev, F. Hayek, J. Jasin etc.
The methods of national welfare resources parametric evaluation are studied in the
works of S. Ajvazjan, G. Becker, N. Zubarevich, I. Maslova, M. Mozhina, R. Nure-
jev, L. Ovcharova, V. Polterovich, J. Rjumina, A. Shevjakov etc. Applicable mecha-
nisms and decision making technologies in the sphere of national welfare resource
management are analyzed in the works of M. Baskova, O. Bogomolov, A. Dynkin, M.
Musin, O. Pchelintsev, S. Rosenfeld, S. Sampler, V. Tretjak, T. Schultz, M. Jagolnit-
ser, etc.
Theoretical analysis of such phenomena as “informational civilization” (R. Abde-
jev, S. Djatlov, M. Kastels, S. Parinov, F. Jansen), “national innovative systems” (K.
Bagrinovskij, M. Bendikov, O. Golincheko, I. Dezhina, J. Lotosh), “intellectual capi-
tal” (E. Brooking, A. Gaponenko, M. Malone, T. Sakaya, L. Edwinsson), “cluster
development strategy” (T. Anderson, A. Weber, M. Iversen, A. Isaksen, N. Kal-
juzhnov, R. Kachalov, J. Christensen, B.-A. Lunvall, A. Ljamzin, L. Markov, N. Na-
grudnaja, P. Nertog, L. Nesta, M. Porter, M. Enright) was also important for paper’s
conception making.
Fig. 1. National welfare as the innovative process rent-yielding factor
Acknowledging high importance of the aforementioned scientists research and not-
ing the fact that there are fundamental approaches for exposing separate facets of the
topic considered in this article, it is necessary to underline, however, that hitherto one
has not realized an approach related to a complex evaluation of the national welfare as
an integrated resource of the nation oriented economic development, one has not ex-
Features of National Welfare Innovative Potential Parametric Indication … 343
posed its innovative-reproductive function within the framework of involving into the
economic system such factors as knowledge and intellect of the nation. Little practical
research has been made with regard to the issue of modernizing mechanisms of its
conversion into innovative economic resources.
Insufficient conceptual-methodological development of the resource approach to
the national welfare analysis in the system of the innovation-oriented economic de-
velopment; resource support of the innovation vector of economic development in
conjunction with its theoretical-applicative topicality determined the purpose of the
research.
2 Intermediate Results of Stages of Research
The purpose of the article is to form a methodological basis and to elaborate a theo-
retical-conceptual model and information-analytical tools system of the national wel-
fare innovation-reproduction function, conditions, mechanisms and implements of
using its resources in the interest of developing an innovative-economic system com-
plex analysis. Achieving the set goal determined the necessity and logical sequence of
solving a set of stage-by-stage theoretical-applicative tasks. The results of fulfilling
said tasks could be formulated as follows:
1. Innovation-oriented development of the present day national economies within
the framework of long-term global evolutional trends is more and more determined by
the national welfare level and its dynamics. The national welfare resources accumula-
tion induces higher volume and quality of human capital, higher labor efficiency,
modernization and efficient innovation-oriented national economies development.
2. The national welfare structure encompasses not only traditional material bene-
fits-resources characteristic for pre-industrial and industrial societies (real monetary
revenue, volume and structure of the personal consumption fund, housing conditions,
employment etc.), moreover, it includes new benefits-resources, having higher mar-
ginal utility (level/quality of education and health of the population, quality of the
social-ecologic habitat, freedom of access to new technologies and scientific discover-
ies, social capital etc.). The definitive result of mentioned factors involving into the
innovative productive cycle is revenue in the form of innovative rent creation which
insures competitiveness of the entire production process.
The economic composition of such national welfare resources conversion is capi-
talization of their competitive advantages in the course of countries integration in the
world market and network world-economic relations, especially, in the high-tech
spheres, based upon high quality of the human capital, in other words, transformation
of said advantages into the source of the added value and into the objects of global
investment activities. Within the framework of the innovation-oriented dynamics,
national welfare thus assumes the function of its resource-factor, increment of which
within the world and state structure of social-economic relations becomes a key pre-
requisite for the innovative economic development trend.
3. Globalization processes exert contradictory influence upon economic mecha-
nisms related to the national welfare resources usage aimed at the support of innova-
tion-oriented development. On one hand, they broaden the innovative-economic space
of the country and the possibilities of converting the resources of its national welfare
344 E. Lazareva
competitive potential into innovative-intellectual resources, on the other, globalization
brings about an additional impetus for bigger inter-state asymmetry, polarization of
the countries innovative development. Macroeconomic indicators analysis character-
ize the level of the countries integration into the global innovation-oriented economy,
which showed a high level of interstate developmental inequity and lack of competi-
tiveness of a set of components of the national welfare. This situation decreases the
level of converting separate components of the national welfare into innovative eco-
nomic development resources.
4. The national welfare resources accumulation-consumption (reserve-flow) values
correlation is a distinctive indicator of the innovative reproduction process cyclical
development. During stagnation, low rate of economic dynamics, accumulated na-
tional welfare is depleted (as the result of mobilizing its certain part in order to insure
innovative economic growth), and during rise, high growth rate, the situation is oppo-
site, national welfare is accumulated due to added national revenue, creating thus an
integrated basis for a long-term incremental trend of social economic innovation-
oriented development.
5. Greater role and larger scale of the economic development innovative factors
change traditional perception of the classic stages of the modern expanded produc-
tion. The stage of accumulating intangible assets - factors of production, which create
the innovative economy resource basis becomes the initial and principal stage in the
new scheme of reproduction economic relations. This stimulates national welfare
accumulation with the view of achieving higher productivity, fist of all, of the intel-
lectual resources, of human capital, creation of the institutional habitat, beneficial for
elaboration and distribution of innovations, and due to these factors higher rate of
innovative economic dynamics.
6. The need to convert of the accumulated tangible and intangible national welfare
resources into innovative development factors is embodied in the new priorities and
strategies of the long-term state economic policy, in accordance with which the inno-
vative growth of economy is due to observance of the principal of correlation and
balance of the imperatives of economic efficiency, social justice and ecologic stability
as the three principal criteria of the innovation-oriented reproductive development of
a high aggregation level. Moreover, this includes the fact that the state innovative
policy assumes a new objective function – the function of balanced social-economic
interests of the national, regional (local) and global economic subjects in the process
of accumulation, reproduction and usage of the national welfare resources with the
view of innovative growth.
Studies of the new model of subject-object relations in the national welfare repro-
duction system and conversion of its elements into resource sources for innovative
growth showed that the “network reality” conditions makes topical the issue of elabo-
ration of a collective strategy for the the national welfare development, in which the
aspect of “co-operation” prevails over the aspect of “competition”, and the classical
model of the civil society, based on legal definitions of liberalism and market regula-
tion, is replaced by the corporate community model (fig. 2).
Features of National Welfare Innovative Potential Parametric Indication … 345
Fig. 2. Subject-object relations’ conceptual model
Coordination and balance of the specific interests of innovation-oriented develop-
ment subjects in reproduction of national welfare as a national benefit, based upon
joint advantages, trust and state-personal partnership make one of the important
methodological principles for the corporate strategy formation, which helps to opti-
mize management and to achieve higher efficiency of national welfare usage with the
view of national economic growth. Determination of the ideal “hierarchical chain” of
interests of the economic subjects and orientation of the adequate stimulation policy
towards it is one of the alternatives for realizing the coordinative methodological
principle.
346 E. Lazareva
7. The national welfare resources inclusion into the economic asset balance of the
country signifies their interpretation as a source for added value in the long-term in-
novative cycle of economic dynamics.
The most adequate approach to economic evaluation of tangible and intangible
components of national welfare is a modified variant of the Hartwick-Solow princi-
ple, in accordance with which it is necessary to consider the innovative rent as the
principal source of the national revenue, a part of which is channeled into national
welfare accumulation, that brings about a higher resource potential of the long-term
innovation-oriented economic development – it is re-invested in better quality of the
human capital and amelioration of the social-ecologic conditions of its reproduction
(education, healthcare, fundamental science, social infrastructure, lower environment
pollution).
8. Rent revenue, which is received due to efficient usage of the new categories of
intangible benefits – informational, innovative, infrastructural, intellectual benefits
that directly insure reproduction of human capital, play a greater role in the national
economies innovation-oriented modernization of national economies. As it is shown
by the analysis of said processes in Russia, taking into account the existing institu-
tional deficits, underdeveloped nature of venture business, lack of a systematic, high-
quality network habitat, favorable for diffusion of innovations and weak interest of
the economic subjects in their elaboration and implementation, capitalization of the
present innovative potential of national welfare (infrastructural, educational-
intellectual, informational welfare) is rather difficult, whereas innovative rent is
gained only in separate, isolated cases. As the result a considerable part of the existing
tangible and intangible national welfare resources, fist of all, intellectual and human
resources, is not capitalized. This fact brings about a lower competitiveness of the
country.
9. Need to indicate and insure elaboration of the national welfare components con-
version into innovative economic resources mechanisms presupposes analysis and
evaluation, to be executed in the state management system, regarding the level and
dynamics of reproducing its four components – quality of the population proper, ma-
terial life standards of the population, quality of the social habitat and quality of eco-
logic state of the natural-economic complex. Complex evaluation of the aforemen-
tioned four components of the national welfare with the view of achieving innovative
economic growth is based methodologically upon usage of sophisticated theoretical-
analytical set of implements including a set of formalized methods and models of
determining latent connections between national welfare and innovation-oriented
economic growth (which form a unity of the innovative reproduction process) as well
as evaluation of the innovative effects due to a higher level of converting different
national welfare components into factors of innovative growth.
The elaborated set of implements allows to analyze efficiency of the existing na-
tional welfare resource structure, to expose its limiting components and to form on
this basis a strategy for a long-term economic policy, aimed at development of institu-
tions, which increase national welfare resources competitiveness and the level of their
conversion into productive sources of innovative economic growth.
A distinctive feature and advantage of the elaborated model set of implements is
the possibility to use it in order to accumulate analytical information regarding the
results and parameters of economic, social, ecologic strategies related to accumulation
Features of National Welfare Innovative Potential Parametric Indication … 347
and increment of the national welfare resources with the view of achieving a higher
national economic dynamics and to thus provide (as opposed to the traditional imple-
ments) a more adequate evaluation of the mechanisms used in the state economic
policies related to support of the innovation-oriented economic development trends
(figure 3) [2].
10. Diagnostics effected on the basis of the set of implements with regard to the na-
tional welfare as an integrated resource of innovation-oriented economic development
of Russia state-of-the-art within the global coordinates framework (figure 4) and
integral innovative effect of its increment showed that due to realized innovative wel-
fare management strategies (including strategies of a higher level/quality of education
and lower sickness rate of the population, higher buying power of its monetary in-
come per person and lower level of poverty, development of the social infrastructure,
higher social-territorial mobility and level/conditions of employment of the popula-
tion, development of smaller business and greater freedom of entrepreneurs, creation
of a dynamic information infrastructure and better access to technologies and science
etc.), the indicators of the Russian economy subjects may be increased approximately
by 1.5 times mostly by means of better social sphere quality [3].
The innovative effect due to national welfare increment and its transformation into
innovative economy resources indicators: level of economic subjects’ innovative ac-
tivities, level of conversion of national welfare into innovative growth competitive
factors, innovative rent capitalization level are the key parameters which characterize
the proportions between accumulation and consumption of the national welfare re-
sources state policy.
The quotients, calculated (with regard to Russia) with the view of proposed struc-
tural model empiric verification and reflecting the dependence between the dynamics
of the innovative activity of the economic subjects parameters and the parameters of
the national welfare resources (average expected lifetime, GDPPPP per person, Gini
index and ecologic stability index) showed that, within the integrated effect indicator
among the four basic components of national welfare, greater importance is held by
the social sphere resources which reflects the priority of social, socially-advantageous
benefits – social capital accumulation sources reproduction. These sources are charac-
terized by such important properties from the point of view of the innovative growth
as: positive network effects and their higher marginal utility in the course of their use;
therefore, the level of the national welfare resources into factor sources of innovative
growth conversion greatly depends upon the state of the social sphere – the elasticity
quotient was 1.724 and the correlation quotient was 0.671, then, following the order
of lower dependence one has the ecologic habitat quality – accordingly 0.463 and
0.324, quality of the population – 0.137 and 0.393 and the material life standards –
0.057 and 0.442.
348 E. Lazareva
Fig. 3. Model set of implements for state strategy’ analytical evaluation
Features of National Welfare Innovative Potential Parametric Indication … 349
Fig. 4. National welfare state-of-the-art estimation
11. In the course of interregional comparison of the conditions and existing limits
for realizing the policy of state national welfare resources reproductive proportions
between accumulation and consumption optimization, one detected a domineering
correlation between the economic subjects innovative activity indicator and indica-
tors of the achieved level of conversion. The calculations effected by the author ac-
cording to special methods showed, in particular, the following typology of de-
pendence of the features of the economic subjects innovative activity on the pa-
rameters of domineering kinds of national welfare resources (social sphere re-
sources) into innovative growth factors conversion which determine the priorities
of the long-term state economic policy: for an economy, which is characterized by
a low, medium, high fully realized dependence of the features of the economic sub-
jects innovative activity upon the parameters of the social sphere resources con-
version, priority belongs, therefore, to the strategy of developing social infrastruc-
ture and higher level/quality of employment of the population, strategies of devel-
oping smaller business and greater freedom of entrepreneurs, strategies of easier
access to scientific achievements and to new technologies, information infrastruc-
ture development (figure 5). The detected innovative effects indicate the priorities
of the social-economic policy, in which the main role belongs to investments into
the innovative national welfare resources: housing conditions, social and informa-
tion infrastructure, science, education, healthcare, culture etc.
350 E. Lazareva
3 Conclusions
The obtained results show that due to social conditions, factors and motives of behav-
ior more important role, social capital resources greater importance, it is necessary to
elaborate a harmonized systematic program of innovation-oriented long-term eco-
nomic policy modernization and to create a favorable social-economic climate in the
country on the basis of the existing national welfare.
The systematic approach means a reconsidered hierarchy of social-economic pri-
orities within the framework of the person oriented innovative economic growth para-
digm. In this light the state has the following tasks of paramount importance: amelio-
ration of the overall conditions of employment and population housing, recreation of
the salary reproductive function (first of all, on the basis of adequate evaluation of the
level/quality of education); accelerated development of the intangible investment
complex and social infrastructure, realization of human-saving social programs; con-
sistent industrial policy which would activate innovative activity mechanisms and
socially responsible behavior of the corporate subjects which are capable of making
their contribution into development of national welfare and human potential of the
nation.
Fig. 5. Spatial strategic “developmental crystal” of Russian national welfare
In the present situation efficient mechanisms of balanced innovation-oriented eco-
nomic development may be formed only on the basis of the state, civil society and
business integrated efforts in order to achieve consistent expansion and rectification
of opportunities for the representatives of different social, professional and territorial
population groups via reproduction of the national welfare resources as a social bene-
fit. This has to be reflected in the system of innovation-oriented long-term social eco-
nomic development strategic management.
Features of National Welfare Innovative Potential Parametric Indication … 351
The national welfare resources will be realized in an efficient way within the
framework of an innovative economy, only if there is a stable need for them from the
reproductive process. The strategic task is to bring about long-term correlation of the
national welfare resources demand and supply in the innovative development of the
economy.
The developed theoretical analytical tool allows to evaluate not only efficiency of
accumulation and usage of the national welfare resources, moreover, it makes possi-
ble to determine the innovative effect due to a higher level of their conversion into the
sources of innovative growth.
References
1. Lazareva, E. I.: National Welfare as an Integrated Resource of Innovation-Oriented Devel-
opment of Economy. Theoretical-Methodological Aspect. Publishing House of the South-
ern Federal University, Rostov-on-Don (2009) (In Russian)
2. Lazareva, E. I.: Strategy of National Welfare in Interest of Innovative Economic Growth
Development: Results of Systematic Parametrical Indication. Economic Newssheet of the
Rostov State University, 3, 65–74 (2010) (In Russian)
3. Aleshin, V. A., Lazareva, E. I.: National Welfare Increment as the Imperative Institutional
Determinant of Regional Systems’ Development in the Innovative Processes’ Globaliza-
tion Context. Social Inequality and Economic Growth, .4, 9–18 (2012)
Matrix Analogues of the Diffie-Hellman Protocol
Alexsander Beletsky1, Anatoly Beletsky1 and Roman Kandyba1
1
Department of Electronics National Aviation University of Kiev,
1, av. Cosmonaut Komarov, 03680, Kiev, Ukraine
alexander.beletsky@gmail.com, abelnau@ukr.net,
romankandyba@mail.ru
Abstract. This paper presents a comparative analysis of several matrix analogs
of the Diffie-Hellman algorithm, namely, Yerosh-Skuratov and Megrelishvili
protocols, as well as alternative protocols based on irreducible polynomials and
primitive Galois or Fibonacci matrices. Binary matrix is primitive, if the se-
quence of its powers in the ring of residues mod 2 forms a sequence of maxi-
mum length ( m sequence). Offer alternative protocols and discuss ways to
improve the reliability of their.
Keywords. Encryption key exchange protocol, the irreducible polynomials, a
primitive element of Galois field, primitive binary matrix
Key terms. Research, CryptographyTheory, MathematicalModelling
1 Introduction
The Diffie-Hellman algorithm (DH-algorithm) [1] assumes that two subscribers –
Alice and Bob both know the public keys p and q , where p is a large prime num-
ber, and q is a primitive root. Subscriber Alice generates a random big number a ,
computes A q a mod p and sends it to Bob. In turn, Bob generates a random big
number b , computes B qb mod p and sends it to Alice. Then subscriber Alice
raises number B received from Bob to her random power a and calculates
Ka B a mod p qba mod p . Subscriber Bob acts similarly, calculating
b ab
Kb A mod p q mod p . It is obvious that both parties receive the same number
K because K a K b . Then Alice and Bob can use this number K as a secret key,
e.g. for symmetric encryption because a foe who intercepts numbers A and B faces
with virtually unsolvable (in a reasonable time) the problem of calculation K , under
the condition, that numbers p , a and b were chosen big enough.
Matrix Analogues of the Diffie-Hellman Protocol 353
2 Yerosh-Skuratov Protocol
In order to form a secret encryption key in the public network by subscribers Alice
and Bob, the authors [2] propose to use DH protocol in the cyclic group of matrices
M , and the matrix M is considered as public information. It is assumed that Alice
generates a random index x , calculates the matrix M x and sends it to Bob. In turn,
Bob generates a random index y , calculates the matrix M y and sends it to Alice.
Then both subscribers raise the matrices obtained from a partner in their secret powers
and calculate the sheared matrix (encryption key) K M xy M yx . The matrix M
must be a high-order matrix (at least 100); so, the authors assert (by the way, without
a proof), cracking key has invincible complexity. However, in [3] it has been proved,
that Yerosh-Skuratov protocol can easily be cracked based on the generalized Chinese
remainder theorem.
3 Megrelishvili Protocol
The essence of the protocol [4] is following. Binary initialization vector V and primi-
tive matrix M of order n are accepted as a public key. Subscriber Alice generates a
random index x , calculates the vector Va V M x and sends it to Bob. In turn, Bob
generates a random index y , calculates the vector Vb V M y and sends it to Alice.
Then Alice computes the key K a Vb M x V M y x , and Bob computes the key
Kb Va M y V M x y . It is quite obvious that using such data exchange protocol,
both parties receive the same private key K , because K a K b K .
The algorithm of generating the matrices in Megrelishvili protocol is fairly simple
and can be explained by the following calculation scheme
1 0 1 0 1
1 0
1 0 1
M 1 1, M 3 1 M1 0 , M 5 0 М3 1 ,
(1)
0 1 0 1 0
0 1 0 1 0
As it follows from (1), the matrices M i , i 1, 2, , are matrices of odd order only
that can cause some difficulties when they are used in cryptography. This shortcom-
ing was remediated by replacing matrices of type (1) by primitive matrices of an arbi-
trary order that is synthesized based on the so-called generalized Gray transforms [5].
The essence of these transforms is explained below.
The matrix form of direct (for simplicity denoted by number 2) and inverse (de-
noted by number 3) classical Gray transforms (codes) [6] can be presented in the form
354 A. Beletsky, A. Beletsky and R. Kandyba
1 1 0 0 1 1 1 1
0 1 1 0 0 1 1 1 (2)
2 : ; 3 : ;
0 0 1 1 0 0 1 1
0 0 0 1 0 0 0 1
where as an example, the order of the matrix n is set n 4 .
Matrices (2), which we call left-sided Gray transform matrices, are in correspon-
dence with the right-sided transform matrix defined by the following relations:
4 : 121 2T ; 5 : 131 3T , (3)
where
0 0 0 1
0 0 1 0 (4)
1 :
0 1 0 0
1 0 0 0
is the matrix (operator) of the inverse permutation.
The set of operators (2) – (4), supplemented by the operator 0, or e (identity ma-
trix), forms a complete set of simple Gray operators. From the elements of simple
Gray operators, one can form so-called composed Gray codes (CGC) generated by the
product of simple (elementary) Gray codes. The simplest examples of CGC 121 and
131 can be seen in (3). Both simple and composed Gray codes have a number of re-
markable properties. Firstly, the corresponding transformation matrices are nonde-
generate and, therefore, are reversible. Secondly, there are simple inversing algo-
rithms for CGC. And, finally, there are “crypto-order” CGC which have the property
of primitiveness. Examples of such codes are given in Tab. 1.
Table 1. Gray Composite codes delivering binary matrices property of primitiveness
The order of the matrix (n)
32 64 128 256
2244424 22533435 2425535 22533435
2442224 22534335 2433534 22534335
12242253 24334225 2435334 24334225
12242443 25224334 22524224 25224334
12252242 222524424 22533334 2222535224
Suppose M is a primitive binary matrix generated by the CGC G . With respect to
such matrices, the following assertion can be easily proved by the test method.
Assertion. The primitiveness of matrices M is invariant to the group of linear
transformations of the CGC G generating matrix M and transformations of
similarity over these matrices.
Matrix Analogues of the Diffie-Hellman Protocol 355
The group includes the following operators: cyclical shift, assess statement,
inversion and conjugation as well as arbitrary combinations of these operators. Trans-
formation forms matrix M p , which is similar to M and determined by the rela-
tion
M p = P M P 1 ,
where P is a permutation matrix.
4 Alternative Protocols
This section proposes two options for alternative matrix protocols of secret key ex-
change on the open channel of communications. The procedure for the formation of
the encryption key K in the first version of the protocol is based on the use of two
public and one private key for both subscribers. As a public key a binary initialization
vector V of n order and any irreducible polynomial (IP) n of n order are chosen.
Private keys are primitive (forming) elements of the Galois field GF (2n ) over the
IP n , from which the subscribers (Alisa and Bob) form the primitive secret trans-
( )
formation matrices Gn a and G( b ) respectively. The element of the field
n
GF (2n ) is primitive over IP n , if the minimum rate e , at which (e 1) mod
assumes the value e 2 n 1 .
Matrix G( n ) we call Galois matrices. The synthesis of algorithm for such matrices
is explained on a concrete example. Let’s IP 8 100101101 , and the generating
element (GE) of subscriber Alisa a 111 . We obtain
1 1 1 1 0 1 1 1
1 1 1 0 1 1 0 1
1 1 1 0 0 0 0 0
.
0 1 1 1 0 0 0 0
A Ga
0 0 1 1 1 0 0 0
0 0 0 1 1 1 0 0 (5)
0 0 0 0 1 1 1 0
0 0 0 0 0 1 1 1
According to (5), the procedure of filling in the matrix Ga is carried out under the
following scheme. First, the GE a is arranged in the bottom row of the matrix. The
elements of this row in the left from the GE elements are filled with zeros. Subse-
quent rows of the matrix (in the direction from bottom to top) are produced by a shift
of previous lines. If left element of shifted line is 0, then the cyclical shift by one bit
to the left (circular scrolling clockwise). In the case where the left element of shifted
line is 1, the conventional shift of the line on one bit to the left and 0 is written to the
vacant right element in line. Digit capacity of these lines is one bit more than the or-
der of the matrix. The vectors corresponding to these lines are given to the residue
356 A. Beletsky, A. Beletsky and R. Kandyba
modulo IP n that returns them the capacity, which coincides with the order of the
matrix n . Subscriber Bob forms similarly the Galois matrix B Gb using his primi-
tive element b .
The introduced Galois matrices have some interesting properties. First, the matrix
product is commutative, i.e. A B B A . At the same time, secondly, if at least one
of the GE is not a primitive of the IP, the commutative property of matrices is lost.
Based on the above properties of Galois matrices a key exchange protocol was pro-
posed.
We consider that initialization vector V and the IP are known. Alice chooses a
secret primitive over GE a , forms a Galois matrix A , calculates the vector
Va V A and sends it to Bob. In turn, the subscriber Bob selects a primitive GE b ,
forms a matrix B that calculates the vector Vb V B and sends it to Alice. After
that, both parties multiply vectors obtained from the partner, in own secret Galois
matrix. Thus, a shared secret key K will be formed by the fact that the product of
primitive Galois matrices over the same IP is commutative, and this implies the
identity
K a Vb A V B A K b Va B V A B .
Instead of Galois matrices G , Fibonacci matrices F can be used in the protocol
with the same success. Fibonacci matrices are associated with Galois matrices by
equation
F G, or F = G ; G F ,
where means the operator of right transposition, i.e. transposition with respect to
the auxiliary diagonal matrix.
In the second alternative embodiment of the protocol the secret key K is com-
puted in two rounds. In the first round, which repeats the above-considered first ver-
sion of the protocol, a common to both subscribers secret binary vector of n th or-
der V p is formed. On the basis of this vector, Alice and Bob compute the common
permutation matrix P . One can propose different ways of constructing matrices P .
Let us consider one of them. Let’s n 8 and N is the decimal equivalent of the vec-
tor V p . The task is to create permutation matrix P8 of order eight for value N .
Choose one or another way of numbering elements of matrices P8 from 0 to 63. Cal-
culate the value n8 N mod 64 and write 1 in that element of the matrix, whose
number is equal n8 . After that, delete from the matrix P8 the row and column, which
contains 1. We obtain a matrix P7 of 7-th order, whose elements are numbered from 0
to 48. Find the value n7 N mod 49 , which is determined by the location 1 of the
matrix P7 and, consequently, in the matrix P8 . Following the proposed method, one
can simply construct a permutation matrix P of any order.
Let proceed to the second variant of the encryption keys protocol. This protocol
uses two public keys, which are the initialization vector V , and the irreducible poly-
Matrix Analogues of the Diffie-Hellman Protocol 357
nomial , and also two private keys. These keys are generated by Alice and Bob as a
random primitive over IP Gees and . The protocol runs in two rounds. In the
first round based on public keys V , and secret GE network operators calculate
the total permutation matrix P . The second round is performed in the following or-
der. Alice chooses a primitive over GE a , forms Galois matrix A , then similar
matrix Ap P A P 1 , computes a vector Va V Ap , and sends it to Bob. In turn,
Bob chooses a primitive over GE b , forms Galois matrix B , then similar ma-
trix B p P B P 1 , computes a vector Vb V B p and sends it to Alice. After that,
both parties multiply vectors obtained from partners on their secret similar Galois
matrix. Thus, the shared key K will be generated due to the fact that the matrices Ap
and B p maintain the properties of primitiveness and commutatively of primary ma-
trices A and B , respectively.
5 Protocol of Vagus Keys
One of the major drawbacks of alternative algorithms key generation algorithms for
open key cipher infrastructure, in particular the mentioned above the way of synthesis
Galois matrix (by the diagonal fill method), is that it could be easily compromised.
To prove that, let's see the vector
Va V G (fn a ) , (6)
created by Alice.
By the theory of polynomials of one variable x , we know that product of any
polynomial n x power of n by x is equivalently either simple shift of polynomial
for one bit left or incrementing the power of polynomial,
x n x n 1 x . (7)
Taking formula (7), let's represent the Galois matrix G (fn a ) the power of n by,
x n 1 x n 1
n2 n2
x x
G (f ) (m od f n ) (m od f ) E , (8)
n n
x x
1
where E the unit matrix.
From formulas (6) and (8) we can get,
358 A. Beletsky, A. Beletsky and R. Kandyba
Va V a (mod f n ) , (9)
where all parts are known, except a . Solving the equation (9), we found:
a Va V 1 mod f n . (10)
For example, let's use the matrix G (fn a ) , given by expression (5), where n 8 ,
a 101101 , f8 101001101 , so f8 is public, a is private keys of protocol. As
initialization vector we choose V 11010010 , that corresponds to invert by modulus
f8 vector V 1 110010 . By formula (9) we get Va 10111111 . Putting the Va and
V 1 is the right side of expression (10) and taking modulus f8 of vectors multiplica-
tion results, enemy (Eva) is getting private key a of Alice. The same way, Eva
could found secret key b of Bob. After secret keys a and b are found it's trivial
to calculate secret key K .
The security of alternative protocols could be increased up to security level of algo-
rithms based on problem of factorization of modular multiplication of big numbers if
we assume that there is secret parameter , both known to Bob and Alice.
The modification of protocol [6] is the be following. Assume, there are authorized
subscribers that have secret parameter as binary vector of n order. Parameter
could be transported from Alice to Bon (or otherwise), e.g. by RSA protocol. Alice is
generating random of n order number a and computing generating element
a a mod f n , (11)
by means of generating element Alice is forming Galois matrix G f a , calculating
n
vector Va V G (fn a ) and sends it to Bob. In the same way, Bob send to Alice vector
Vb V G (fb ) , where b b mod f n .
n
As it shown above, generating elements a and b could be easily computed, so
authorized subscribers Alice and Bob, but not Eva, could calculate secret parameter
of partner. As example, by formula (11) Bob calculates a a 1 mod f n , that
gives him and Alice ability to calculate secret key K a b mod f n . Key K as
well as any function of it, could be taken as a secret parameter K for session
key generation for public key cipher channels.
We call that way of key generation – protocol (algorithm) of vagus keys. Vagus
keys algorithm could be used in both motioned above protocols. The major benefit of
vagus key generation algorithm is protection from "man in a middle" type of attack.
It's been archived by including in Galois matrices key generation elements of secret
element , known only by Bob and Alice. In case of secret element is changed
Matrix Analogues of the Diffie-Hellman Protocol 359
by element e of Eva, makes it impossible to Eva to calculate parameters a , b as
well as general cipher key K .
6 Conclusions
The article analyzes the known matrix algorithms for exchanging encryption keys
between subscribers of a network of open communication channels. The algorithms
are based on the modified asymmetric Diffie-Hellman protocol. The essence of the
modification is reduced to replacing the large prime numbers of Diffie-Hellman algo-
rithm by assurance nondegenerate primitive binary matrices of high order. Methods of
synthesis of these matrices are proposed based on both the generalized Gray codes,
and irreducible polynomials. New key exchange matrix protocols have been devel-
oped. The protocols developed are superior for cryptographic strength to known cryp-
tographic protocols, particularly Yerosh-Skuratov and Megrelishvili protocols de-
scribed in this paper.
The proposed variants of vector-matrix protocols for exchanging by cryptographic
keys on open communication channels have a good prospect to be applied for sym-
metric encryption in computer networks protected from the substitution of data, pro-
viding the necessary level of protection of private keys from unauthorized access.
These protocols can make a strong competition to more resource-intensive RSA pro-
tocol.
References
1. Diffie, W., Hellman, M. E.: New Directions in Cryptography.IEEE Transactions on Infor-
mation Theory, IT-22(6), 644–654 (1976)
2. Eros, I. L., Skuratov, V. V.: Addressing Message Transmitting Using Matrices Over GF
(2). Problems of Information Security. Computer Systems, 1, 72–78 (2004) (In Russian)
3. Rostovtsev, A. G.: On the Matrix Encryption (Criticism Yerosh-Skuratov Cryptosystem),
http: www.ssl.stu.neva.ru/psw/crypto/rostovtsev/Erosh_Skuratov.pdf (In Russian)
4. Megrelishvili, R. P., Chelidze, M. A., Besiashvili, G. M.: Unidirectional Matrix Function -
High-Speed Diffie – Hellman’s Analog. In: Proc. 7-th Int. Conf. Іnternet - Education - Sci-
ence 2010. VNTU, Vіnnitsya, 341–344 (2010) (In Russian)
5. Beletsky, A. Ja., Beletsky, A. A., Beletsky, E. A.: Gray Transformations.V.1. Fundamen-
tals of the theory. V. 2. Applied aspects. NAU Publishing House, Kiev (2007) (In Russian)
6. Beletsky, A. Y., Beletsky, A. A.: Synthesis of Primitive Matrices over a Finite Galois
Fields and their Applications. Information Technology in Education: Collected Works, 13.
Kherson: KSU, 23–43 (2012) (In Russian)
Are Securities Secure: Study of the Influence of the
International Debt Securities on the Economic Growth
Darya Bonda1 and Sergey Mazol2
1
Belarus State Economic University, Minsk, Belarus
bondadasha@gmail.com
2
Academy of Public Administration, Minsk, Belarus
mazols@yandex.ru
Abstract. The paper studies the interdependence of the amount of international
debt securities, amounts outstanding by country (borrowers) and the GDP
growth by country. The author have chosen 34 countries, that represent every
region included in the BIS classification, that is developed countries, offshore
centers, developing Europe, Latin America, Asia and the Pacific and Africa. It
was found that the excessive amount of such type of securities in comparison
with GDP leads to slowdown in the economic growth next year.
Keywords. International debt securities, Economic growth, Financial crisis
Key terms. Development, MathematicalModel
1 Introduction
The last financial turmoil has revealed the drawbacks of the existing global financial
system. Surprisingly, the worst crisis since the Great Depression has offered a range
of opportunities to the world society: to examine the system, exclude “toxic” elements
and introduce new methodology to financial regulation.
During the last decades new avenues for financing were creating, deepening the fi-
nancial system aside from widening the choice of monetary instruments [1] that have
caused overestimation of assets and, consequently, financial collapse.
Despite this issue is under thorough control of Bank of International Settlements,
Securities and Exchange Committee, International Derivative and Swap Association,
Securities Industry and Financial Market Association, every scientists, analyst, gover-
nor, outstanding person and a regular student has its own interpretation of how the
crisis works, its causes and consequences.
Are securities secure … 361
One of the reasons for the growing financial instability was the excessive amount of
various types of securities both in national economies and international arena as well
as the complexity of the securities issued. New types of financial instruments usually
at first are accepted as great invention of humanity, then, especially during recessions
are usually blamed for crisis for speculation reasons [7]. After the recovery, they are
still widely spread all over the world. Futures, options and other derivatives have
experienced such an attitude [4]. International debt securities are considered to be a
financial instrument. The amount of securities outstanding in 2007, i.e. country’s
liabilities, could prevent countries from sustainable growth in 2008.
To the author’s point of view, it is reasonable to study the interdependence of the
amount of International debt securities, amounts outstanding by country (borrowers)
and the GDP growth by country. The presence of such interdependence can allow us
to criticize this type of securities and advise the countries to minimize their usage for
the sake of sustainable economic growth.
2 Results
The authors are analyzing the interdependence of the amount of international debt
securities outstanding in 2007, and the economic growth, expressed in GDP index in
the research.
Debt security is a negotiable financial instrument serving as evidence of debt [5].
The statistics on international debt securities issues cover long-term bonds, notes,
short-term money instruments [2]. Debt securities include government bonds, corpo-
rate bonds, CDs, municipal bonds, preferred stock and collateralized securities (such
as CDOs, CMOs, GNMAs). Debt securities may be protected by collateral or may be
unsecured, which underlines the importance of scrutinizing them as one of the key
instruments of securitization. Collateralized debt obligations are considered to be a
risky instrument as far as their coupons and principal repayments are dependent on a
diversified pool of loan and bond instruments, either purchased in the secondary mar-
ket or from the balance sheet of an original asset owner (Handbook of Securities Sta-
tistics). Consequently, through assessing the value of underlying assets, collateralized
debt obligation as well as other asset-backed securities spread risk while diversifying
it, meanwhile creating a range of credit derivatives. These instruments are widely
used to make the debt more liquid and make the money lent work as if they were not
borrowed and, furthermore, get a margin. Therefore, the author finds it crucial to pay
significant attention to this kind of financial instrument as a mean of spreading risk of
insolvency of an entity within the international scale.
The BIS definition of international securities (as opposed to domestic) is based on
three major characteristics of the securities: the location of the transaction, the cur-
rency of issuance and the residence of the issuer. International issues comprise all
foreign currency issues by residents and non-resident in a given country and all do-
mestic currency issues launched in the domestic market by non-residents [2].
GDP in current price, purchasing power parity, is the second element of the re-
search. It was chosen as an indicator of the national output, combining real and finan-
362 D. Bonda and S. Mazol
cial sector, thus reflecting the size of the economy. The amount of international debt
securities can be compared to the GDP, as both indicators reveal the capacity of the
countries’ economies.
In the figure 1 one can see historical correlation between the international debt se-
curities and GDP in current prices (data from [1], [2], [3], [8]), which shows that the
interdependence between 2 components really exists, in addition, during 2002-2008
the line shows different slope. In 2004 and 2005 the inclination is lower, which means
the slowdown in GDP growth rate and IDS amounts and, on the contrary, in 2007 and
2008 the graph indicates the rise of the world economies.
Historical Correlation between IDS and GDP, world, 2002-2008
70000
60000
50000
GDP, bln UDS
40000
30000
20000
10000
0
0 5000 10000 15000 20000 25000 30000
IDS, bln USD
Fig. 1. Historical correlation between IDS and GDP, world 2002-2008
As mentioned in the title, the graph shows the world’s tendency. The last economic
crisis has damaged major economies leaving some small and developing economies
untouched, despite of its scale. This means that by-country analysis is necessary to
provide the real evidence of such correlation.
The authors have chosen 34 countries that represent every region included in the
BIS classification that is developed countries, offshore centers, developing Europe,
Latin America, Asia and the Pacific and Africa. Although the most variation in the
amount of securities outstanding had been noticed while scrutinizing the data from
developed countries the data utilized represents each continent. Africa is represented
by Egypt, Lebanon, Saudi Arabia and UAE, Asia and the Pacific – the Philippines,
Singapore, Japan and China, Developing Latin America – Argentina, Brazil, Colom-
bia, Costa Rica, Developing Europe – Belarus, Bulgaria, Czech Republic, Estonia,
Russia, Ukraine, offshore centers – by the Bahamas, Developed economies – by Aus-
tralia, Austria, Belgium, Canada, Finland, France, Germany, Greece, Iceland, Italy,
Norway, Spain, Sweden, the UK and US. The X, independent variable, is the amount
of international debt securities by country outstanding in 2007 divided by the nominal
GDP in billion of USD in 2007, whereas the dependent variable is GDP growth rate
in 2008 in comparison with 2007. The linear regression is shown on figure 2.
Are securities secure … 363
Correlation between IDS/GDP 2007 and GDP growth rate
2008/2007
y = -10,071x + 20,058
R2 = 0,5193
35
GDP growth rate 2008/2007
30
25
20
15
10
5
0
-50,00 0,50 1,00 1,50 2,00 2,50 3,00 3,50
-10
-15
IDS/GDP coefficient
Fig. 2. Correlation between IDS/GDP 2007 and GDP growth rate 2008/2009
The model is described by formula 1:
GDP= 20,058-10,071*COEF (1)
where GDP – nominal growth rate in 2008 compared to 2007;
COEF – IDS/GDP coefficient as a ratio of IDS amounts outstanding 2007 and GDP
(in current prices, power of purchasing parity, 2007)
The slope is negative, which means the opposite correlation between variables, the
bigger the coefficient in 2007, the lower the GDP growth rate in 2008, which makes
sense and confirms the author’s theory. The correlation coefficient, which shows the
fraction of relationship between variables is -0,72 or 72% of opposite relationship.
Therefore, the relation is considered to be strong. The determination coefficient is
0,52 or 52%, which means the variation of the dependent variable is explained by
52% by the independent one.
The linear regression has been chosen because the goal of the author was to find the
existence of the relationship between variables, and in addition, negative one, how-
ever, admitting the complexity of the relationship, not exactly finding the most appro-
priate one.
Among the obstacles which prevent linear model from being “best fit” are, first of
all, the different level of financial market development and, thus, the vital need for
specific financial instruments usage. For example, Belarus is not yet ready for devel-
oping credit derivatives, besides, the amount of the securities outstanding has been the
same for a couple of years, that means that the amount of securities is not the princi-
pal reason for economic crisis. Another reason is overstated prices in some countries
which result in the high inflation level, which increases the GDP index far too high to
depict the precise relationship between variables. Examples can be Ukraine, Russia
and Argentina. As far as offshore centers are concerned, their GDP doesn’t always
364 D. Bonda and S. Mazol
illustrate the real production but the value of financial operations, so they need indi-
vidual approach. However, there are some deviations in the model: for example, Ice-
land – at the lowest point, because of negative GDP growth, or the US and The UK
with the lowest GDP growth in 2008/2007 and highest amount of securities within
examined countries.
In the real economic world one can hardly obtain “pareto efficient” outcomes, thus
no sector of the economy can be better off without making another worse off [6].
Therefore, the increase of the international debt securities, i.e. rise in liquid liabilities,
results in lowering the GDP growth rates. Even though the GDP measures only mar-
ket production and cannot totally used as a measure of country’s well-being [6], a
better measurement of economic activity hasn’t been offered yet. Thus, the influence
of financial sector on the overall economy proves the existence of “systemic external-
ity” [6] of the financial market.
Ranging Countries
35 3,50
30 3,00
IDS/GDP COEFFICIENT 2007
Nominal GDP growth rate
25 2,50
20 2,00
15 1,50
10 1,00
5 0,50
0 0,00
Brazil
Sweden
Czech
US
UAE
Iceland
Ukraine
Lebanon
Canada
Japan
France
Finland
Austria
Spain
Singapore
Greece
Australia
Colombia
Estonia
Costa rica
China
Saudi Arabia
Bulgaria
Argentina
Ukraine
Russia
Belgium
Bahamas
Norway
Belarus
Philippines
Germany
Italy
Egypt
-5 -0,50
Fig. 3. Ranging countries
The figure 3 is built with the purpose to prove the main idea of the research: the
bigger the volume of IDS outstanding (here IDS/GDP coefficient to make it adjust-
able to different economy sizes), the lower the GDP growth rate. Though, the graph
doesn’t totally reflect the inverse dependence, it shows the tendency and proves the
idea for the majority of countries: while the slope of the GDP growth rate curve is
positive, the slope IDS/GDP coefficient curve is negative.
Among the factors that prevent the ideal illustration of the theory is the nominal
character of the variables. One of the elements is the security market capitalization
which can not possibly be turned into real one, therefore the usage of nominal GDP
growth rate is reasonable. Another factor is the uniqueness of economic development
of the countries, like the inflation and unemployment level and the unique interde-
pendence of the economic sectors. Consequently, it is necessary for the countries to
develop its own targeting rules, using its own IDS/GDP ratio and work out the model
of the development that will be adequate for a single economy specifications.
Are securities secure … 365
Perhaps, it is quite possible to work out the limit in IDS/GDP ratio which countries
shouldn’t exceed in order not to affect sustainable development, but this work re-
quires, first of all, individual approach and demands profound knowledge on eco-
nomic development of each country.
3 Conclusions
The overall significance of the model is to show the impact of the issuance of interna-
tional debt securities on the economic growth of the countries. Since the slope of the
regression line is negative, the excessive amount of such type of securities in com-
parison with GDP leads to slowdown in the economic growth next year. Due to the
fact that the issuance of securities is partly managed by governments and financial
organizations (in case of the US – Securities and Exchange Committee (SEC), for
example), the issuance of them can and should be regulated. After the first wave of
financial crisis has gone some rating agencies, for example, S&P have offered down-
grading system for some risky instruments, thus protecting the market from distribut-
ing of excessive risk, as well as SEC has been promoting a new bill to the White
House for a while (Reuters). This information shows that global financial society is
already trying to react and eliminate some of the causes of the financial crisis, how-
ever, not yet fruitful.
The global financial crisis has proved the necessity of the permanent monitoring
and control of the financial system, especially with regard to financial architecture
and innovations [6]. The research has revealed that the volume of the international
debt securities should be a subject of the country’s guideline to optimal control and
efficient rules for financial policy. Increase in the IDS/GDP ratio may lead to finan-
cial instability and increased involvement in the global financial market. In addition,
secure financial instrument usage with careful risk control limits allows financial and,
therefore, the whole economic system to reach sustainable development.
References
1. Bank for International Settlements: Financial Market Developments and their Implications
for Monetary Policy. №39. BIS, Basel (2008)
2. Bank for International Settlements: BIS Guide of Securities Statistics. №39. BIS, Basel
(2009)
3. Bank for International Settlements: BIS Quaterly Review, December 2009. BIS, Basel
(2009)
4. Chibrikov, V.: History of the Derivatives. Economic Issues, 3 (2008)
5. International Swaps and Derivatives Association: Handbook on Securities Statistics, ISDA
(2007)
6. Stiglitz J.: Regulation and Failure in New Perspectives on Regulation, In: The Tobin Pro-
ject, pp. 11–24, Cambridge (2009)
7. Suetin A.: About the Reasons for Financial Crisis. Economic Issues, 1 (2009)
8. World Bank Database, www.worldbank.org
How to Make High-tech Industry Highly Developed?
Effective Model of National R&D Investment Policy
Oksana Moiseeva1 and Sergey Mazol2
1
Belarus State Economic University, Minsk, Belarus
oksuta13@mail.ru
2
Academy of Public Administration, Minsk, Belarus
mazols@yandex.ru
Abstract. The paper validates the relations between the share of public and pri-
vate R&D spending and the effectiveness of national R&D sector. It states that
in order to implement effective and profitable “high-tech policy”, governments
have to intensify the share of business sector in Gross Domestic Expenditures
on R&D. At the same time it is necessary to preserve the definite “government
share” in R&D investments, as reduction of it up to certain extent gives the
negative effect.
Keywords. High-tech industry, R&D investment policy, Private v. Public R&D
Spending, Exports of High Technology Products
Key terms. Development, MathematicalModel
1 Introduction
It is impossible to deny that people all over the world benefit from new technologies
which lead to healthier lives, greater social freedoms, increased knowledge and more
productive livelihood. Each day sees additions to the literature, much of which in-
cludes reports on the establishment or expansion of R&D facilities and programs that
are designed to take the best advantage of highly qualified resources. Nowadays there
are practically no governments and politicians that would miss a chance to stress the
importance of innovations in economy. According to the judgments of some experts,
GDP growth of developed countries up to 50-90% is determined by technological
progress and innovations [7]. The developing countries, in their turn, extremely need
competitive high-tech industry, not only because being usually one of the most profit-
making and cost-efficient industries it contributes to economic prosperity by itself,
but also because technological achievements give them a chance to promote and make
competitive on the global arena all other economic sectors, narrowing in this way the
economic gap between the highly-developed countries and the developing ones.
How to Make High-tech Industry Highly Developed? … 367
Still the results of R&D policy in the countries of post soviet space frequently leave
much to be desired. The most prominent achievements in the sphere of industrial
R&D belong to the most developed countries such as USA, Japan, European Union.
Current literature is replete with reports on the expanding R&D activities in China,
India, South Korea and Singapore. Meanwhile Belarus, Ukraine, Uzbekistan,
Moldova still cannot boast prominent commercial achievements in R&D. That’s why
this research paper aims at analyzing and investigating of those factors and incentives
that turn national innovative efforts, resources and potential into visible and profitable
high-tech results.
Research and development (R&D) comprise creative work undertaken on a system-
atic basis in order to increase the stock of knowledge, including knowledge of man,
culture and society, and the use of this stock of knowledge to devise new applications
[2].
The UNESCO Institute for Statistics claims that the clearest trend in global R&D
activity between 1996 and 2005 was the increasing percentage of GDP devoted by
countries all over the world to R&D (R&D intensity has more than doubled in 9% of
the countries surveyed, including China, Thailand, Tunisia and others; in 48 out of 89
countries surveyed the percentage of GDP devoted to R&D has significantly in-
creased) [4]. Surely, sustained R&D investment is a key to economic growth. But
those are strong words that are easy to follow in good economic times, but more diffi-
cult to follow in bad economic times. R&D expenditures are among the first to be cut
during recessions. Preliminary data (official statistics on R&D are available only until
2007) suggest that companies have reduced their R&D investment in the aftermath of
the crisis. In 2008 the industrial companies despite the challenging economic times
continued growing their R&D budgets, expanding by nearly 6,1%, or more than $60
billion, from what they spent in 2007. Despite their good intentions, when the down-
turn turned from mild to severe, industrial firms were forced to cut their R&D budg-
ets. Total industrial R&D spending dropped by 1% or nearly $10 billion overall in
2009 from what was spent in 2008 [6]. These findings are consistent with historical
trends showing that R&D expenditure exhibits larger variations than gross domestic
product (GDP) over the business cycle. Hence, any drop in GDP would result in an
even larger decrease in R&D expenditure [3].
The 2010 Global R&D Forecast, created by Battelle analysts and the editors of
R&D Magazine, predicts overall global R&D will increase 4.0% in 2010 to $1,156.5
billion from $1,112.5 billion spent in 2009. This increase will mostly be driven by
continued spending by China and India, who will drive a 7.5% increase in Asian
R&D. American R&D spending is expected to increase 3.2% to $452.8 billion, while
EC spending will only increase 0.5% to $268.5 billion in 2010. This forecast espe-
cially stresses a trend of falling the spending of both the Americas (U.S., Canada,
Mexico, Brazil, and Argentina) and the EU behind the spending levels seen in Asian
countries (India and China). Even Japan, the 2nd largest R&D spender in the world, is
now trailing the level of spending by China and India [6].
It is really hard to measure innovations, as its manifestation within the economy is
larger and more complex than what one indicator or index can capture and reflect.
Many aspects of technology creation, diffusion and human skills are hard to quantify.
368 O. Moiseeva and S. Mazol
Still in order to estimate nation’s technological achievements and the level of innova-
tive progress it is possible to use a great variety of indicators. The most frequently
used ones are the following:
The number of patents granted to residents (per million people), the number of new
trademarks
Receipts of royalty and license fees (US$ per person)
The number of researchers in R&D (per million people or per thousand employees)
Population with tertiary education and youth education achievement level, new
science and engineering (S&E) graduates per 1000 population
Science and engineering degrees (% of all new degrees)
% of firms with new-to-market product innovations (as % of all firms)
Sales and exports of high technology products and many others
However, most of the indicators mentioned above describe only the quantitative
side of innovation process, but not the efficiency of national R&D investment policy.
Furthermore, some of these indicators are not representative due to considerable legis-
lation differences among the countries (for example, low patenting activity in India
and China is explained mostly by underdeveloped system of intellectual property
rights’ protection than by lack of innovations) [3]. That’s why this research paper
concentrates mainly on the share of high-technology exports as % of manufactured
exports as the most representative indicator of competitive commercialization of na-
tional scientific researches (Y-variable).
While speaking about the factors of successful innovation policy, it is important to
remember that there are no ideal models in complex economic systems, and each
economic or social parameter is subjected to multitude of different impacts and fac-
tors. As regards national high-tech development, it is affected by such economic con-
ditions as:
International openness to trade and investments (assessed by such indicators as
export and import ratio to GDP (Exp+Imp/GDP*100%), the level of trade
weighted average tariff)
The commitment to market values and developed market economy infrastructure
The accessibility to the venture financing for start-ups
The competitiveness of national economy on the global arena
The volume of R&D spending (in billions of $)
Industry innovation expenditures
The influx of direct foreign investments and many others
Public and private R&D expenditures (% of GDP)
Research and development (R&D) expenditure is one of the most widely used
measures of the innovative efforts of firms and countries. Most surveys devoted to the
technological achievements of the countries concentrate the attention mainly on the
level of R&D intensity of this country as the main factor. But while R&D as a percent
of GDP figures are bandied about as indicators of the strength of the national com-
mitment to scientific research, they have relatively little meaning in terms of just how
that investment contributes to the growth and welfare of the country.
The more important data are those that tell you who is providing the funding, who
is doing the work, how the money is being spent, and what the priorities, thrusts, and
How to Make High-tech Industry Highly Developed? … 369
directions are. In brief, it is the internal structure of the R&D enterprise and the roles
and interplays among the different sectors that have a bearing on the manner in which
the investment in R&D has the desired societal benefit outcomes of economic secu-
rity, improved health care, and the like. The R&D expenditure is generally broken
down among 4 sectors: business enterprise, government, higher education and private
non-profit institutions. In this research the share of business financed R&D was se-
lected for thorough econometrical analysis (X variable).
2 Results
The basic hypothesis based on the preliminary insights into the statistical data sug-
gests that, in order to implement effective and profitable “high-tech policy”, govern-
ments have to intensify the share of business sector in GERD (Gross Domestic Ex-
penditures on R&D). But at the same time it is necessary to preserve the definite
“government share” in R&D investments, as reduction of it up to certain extent gives
the negative effect.
Figure 1 illustrates the average indications of high-technology exports (red line) and
business expenditures on R&D (blue line) during the period 2000-2005 for 20 coun-
tries (the countries were arranged in order of business R&D share extension) [1], [3],
[7].
We can see from the graph that the supposed rule is valid for the countries disposed
in the range of 0-60 % share of business sources in R&D expenditures: the higher
share of business sector in R&D means the higher indications of high-tech export.
Fig. 1. The average indications of high-technology exports and business expenditures on R&D.
Business R&D – % of Gross Domestic Expenditure on R&D financed by Business Sector;
HTexport – % of high-technology export in total manufactured export
However we cannot fully rely on average indications, as each country has its own
peculiarities, historic and geographic conditions and many other factors that can de-
termine high-technological specialization of export. Moreover, here only 20 countries
were taken into account.
370 O. Moiseeva and S. Mazol
That’s why it is more interesting and important to examine the changes in the share
of high-technology exports depending on the changes in the structure of R&D financ-
ing.
The Model description:
X (∆privR&D) – shift in the share of business sector in R&D financing. Business
R&D expenditures as % of total R&D expenditures – the indicator reflects the per-
centage of total investment in research and development originating from the business
sector;
Y (∆HTexp) – shift in the share of high-technology exports as % of total manufac-
tured export during the period 2000-2005.
High technology export is exports of products with a high intensity of research and
development. They include high-technology products such as those used in aerospace,
computers, pharmaceuticals, scientific instruments and electrical machinery [4].
The statistical analysis of the data across 63 countries in the world for the period of
2000-2006 intended to reveal the correlation between changes in R&D expenditures
structure and export structure. It will be studied linear regression.
According to the statistical analysis it was revealed the following general tendency:
the share of the high-tech export increases in most of the countries surveyed along
with the growth of business financed share in R&D investments.
The econometric model on the basis of the statistical data has the following outlook
(figure 2):
∆HTexp= –1,8071+0,7129*∆privR&D (R2=0,636),
which means the increase of high-technology export share by nearly 0,7% if the share
of private R&D expenditures grows by 1%.
Fig. 2. The linear regression
With the purpose of further analysis the countries were classified into 3 categories:
The countries with traditionally high share of private sector in R&D (>60%). This
group includes such countries as Belgium, Denmark, Finland, Germany, Ireland, Is-
rael, Japan, China, Luxembourg, Switzerland, USA (figure 3).
How to Make High-tech Industry Highly Developed? … 371
Fig. 3 The countries with high share of private sector in R&D
The countries with medium share of private R&D expenditures in the range from
40% to 60% are composed of such countries as Austria, Brazil, Croatia, Cuba, Czech
Republic, France, Hungary, Netherlands, Spain, Great Britain (figure 4).
Fig. 4 The countries with medium share of private sector in R&D
The countries with traditionally low share of private sector in R&D (<40%) are
such countries as Azerbaijan, Belarus, Bulgaria, India, Iran, Latvia, Pakistan, Poland,
Portugal, Russia, Ukraine.
Fig. 5 The countries with low share of private sector in R&D
3 Conclusions
1. For 3 groups of countries (with different level of business expenditures) the trend
line has the positive angle, which confirms the basic hypothesis. The peculiarity
here is that the elasticity of high-tech exports to the private R&D investments is
higher for the 2nd group of countries (where 40-60% of R&D is financed by busi-
ness sector). It is the diapason in which the most drastic changes in export struc-
ture happen with the increase or decrease of business share in R&D investments.
372 O. Moiseeva and S. Mazol
Also it is important to highlight that the 3rd group of countries (with low participa-
tion of business sector) mainly has the tendency to declining share of high-tech ex-
port (most of the countries are located in the 3rd quadrant on the graph).
2. The share of business investments in R&D to the extent more than 80% may cause
the decline in high-tech export. So the government expenditures are an important
factor of accelerating the further R&D investments. According to the Harrod–
Domar theory more investment leads to capital accumulation, which generates
economic growth. Regarding the R&D investment policy, it is possible to make an
assumption that expenditures of the government on R&D create the basis and in-
dispensable minimum from which further R&D investment activity is multiplied.
3. Time lag according to statistical data is no more than 1 year.
Explanation of the results, received in the research.
The government is involved mainly in financing the fundamental investigations and
basic research (basic research is experimental or theoretical work undertaken primar-
ily to acquire new knowledge of the underlying foundation of phenomena and observ-
able facts, without any particular application or use in view [2]), which implies low
degree of commercializing of that kind of R&D.
The main commercial projects of the government in R&D sphere are concentrated
in such fields as defense, healthcare, space programs, infrastructure. The governments
of the developing countries invest in import-substituting industries, which also mean
low level of transforming the financed R&D into high-technology export.
Finally, even in the cases when the government takes the lead in innovation financ-
ing and implements different governmental programs for innovative development, it
cannot respond better to the changing market necessities and conditions, than private
investors and companies that are interested to the maximum extend in the commercial
success of their investigations. The governmental programs on the other side fre-
quently tend just to expand the range of goods, but not the technological structure of
industry and its qualitative parameters.
Policy recommendations.
It is important to focus on increasing efficiency in R&D spending rather than meet-
ing a specific spending level. The efficiency and competitiveness of R&D investment
policy, in its turn, can be achieved by expanding of the role of business sector in R&D
financing up to 70-80% (it is important to use economic incentives such as tax ex-
emptions, for example)
It is necessary to preserve the government spending on R&D within the level of 15-
20% of the total expenditures, maintaining at the same time flexibility in allocating
public R&D funds. The government should concentrate mainly in the basic researches
sphere and accelerate in this way other fields of R&D investments.
References
1. OECD: Main Science and Technology Indicators, http://www.oecd-ilibrary.org/science-
and-technology/main-science-and-technology-indicators/volume-2008/issue-2_msti-v2008
-2-en-fr;jsessionid=2obsve2k0wjt3.x-oecd-live-01(2008)
How to Make High-tech Industry Highly Developed? … 373
2. OECD: Factbook 2008: Economic, Environmental and Social Statistics, http://www.oecd-
ilibrary.org/economics/oecd-factbook-2008/world-population_factbook-2008-graph1-en
(2008)
3. OECD: Science, Technology and Industry Scoreboard, http://www.oecd-
ilibrary.org/content/book/sti_scoreboard-2009-en (2009)
4. UNDP: Human Development Report 2007/2008,
http://hdr.undp.org/en/reports/global/hdr2007-2008 (2007)
5. Battelle and R&D Magazine: Global R&D report 2008, http://www.asiaing.com/2008-
global-r-d-report.html (2008)
6. Battelle and R&D Magazine 2010: global R&D funding forecast,
http://www.rdmag.com/topics/global-r-d-funding-forecast?page=2 (2010)
7. World Bank database, www.worldbank.org
Econometric Analysis on the Site “Lesson Pulse”
Alexander J. Weissblut
Kherson State University, 1, 40 rokiv Zhovtnya Street, 73000, Kherson, Ukraine
veits@ksu.ks.ua
Abstract. In this article the site “Lesson pulse” is considered, as the tool allow-
ing the teacher to receive the objective information on a course and results of a
lesson in a mode online. However adequate interpretation for results of such in-
terrogations is impossible, while we will not separate true students from others.
Besides, interpretation for results of interrogations and the decision-making,
grounded on it, demands to realize, what exactly this concrete group means by
clearness of an explanation, objectivity of the marks etc. For anonymous inter-
rogations it means necessity of correlation and regression analysis for their re-
sults and an estimation of the statistical significance of the received results. So
it means necessity of econometric analysis.
Keywords. Factor, statistical, econometric, analysis, correlation, decision-
making
Key terms. Research, Management, Model, KnowledgeManagementProcess,
KnowledgeManagementMethodology, MathematicalModeling
1 Introduction
In this article the site “Lesson pulse” is considered, as the tool allowing the teacher to
receive the objective information on a course and results of a lesson in a mode online.
It allows for the student or the pupil at any moment to react to a lesson course, having
answered one or several questions, for example:
1. Is it interesting to you at a lesson?
2. Is it accessible (clear) an explanation?
3. Are you tired? Whether arranges you the rate of an explanation?
4. There are at you questions to the teacher?
5. Whether the marks are objective?
(Formulations of questions are defined by the teacher). As a result of an average of
these responses the site produces on the monitor screen the data about a lesson state,
its "pulse" in a mode online. At any moment the teacher can ask to answer all simul-
taneously such or more profound groups of questions (their examples are given be-
low). So, he can measure the “lesson pulse” just at this moment. Such interrogations
Econometric Analysis on the Site “Lesson Pulse” 375
do not demand computer auditorium by all means: they can be carried out on one
tablet, and then results can be transferred to a site.
However adequate interpretation for results of such interrogations is impossible,
while we will not separate true students, for which educational process is a consider-
able part of their life, from those, who would prefer to keep far away from it. Besides,
interpretation for results of interrogations and the decision-making, grounded on it,
demands to realize, what exactly this concrete group means by clearness of an expla-
nation, lesson atmosphere, objectivity of the marks etc. For anonymous interrogations
it means necessity of correlation and regression analysis for their results and an esti-
mation of the statistical significance of the received results. So it means necessity of
econometric analysis.
1. All groups of questions considered further have been chosen in result of "brain-
storming" where students of fourth year study of the Faculty of physics, mathemat-
ics and informatics at the Kherson state university acted as experts. This expert in-
terrogation has been constructed by a technique of " six hats of thinking ” E. Bono
[1], which provides the maximal openness and relaxedness of participants. In all
cases the opinion has unanimously been expressed, that the given set of questions
is full and fair.
2. Then students of speciailties “physics”, “mathematics”, “informatics” and “pro-
gram engineering” of the Kherson state university have been interviewed under
such essential requisites. The respondents estimate each question from 0 (at firm
“no”) up to 10 (at firm “yes”). He arbitrarily sets a name of the folder containing
his interrogation (i.e. his key). The volunteer – a participant of interrogation – col-
lects all folders in one main folder and sorts them here (i.e. shuffles). Only after
that the main folder was transferred to the teacher: this simple and open procedure
guaranteed to participants anonymity of interrogation. Alternative and technically
simpler variants are answers to the site and to a tablet: the variant choice is defined
by a kind of interrogation and level of trust of an audience to the interviewing
teacher.
3. Results of interrogation then are transferred to a site “Lesson pulse”, which is real-
ized in language PHP and uses database MySQL (see [2]). The queries realizing
now on the site give out results of the econometric analysis of interrogation. They
include the plural correlation analysis of factors and an estimation of the statistical
importance of the received results with use of criteria Student and Fisher (see [3]).
The site interface is oriented to the user, generally speaking, nothing knowing
about the econometric analysis.
2 The Analysis of Interrogations on Results of Lesson
and Feedback
Results of interrogation about lesson and interrogation Feedback are, of course, abso-
lutely various depending on a lesson, a teacher, an audience etc. However the correla-
376 A. J. Weissblut
tion analysis of factors led to similar outcomes (at 20% a significance level by crite-
rion of Student). Everywhere below we use the interrogations of 421 groups (special-
ity “mathematician”), having typical species (fig. 1).
1
0,8
0,6
0,4
0,2
0
1 2 3 4 5 6 7 8 9 10 11 12 13
-0,2
-0,4
-0,6
Fig. 1. Histogram for distribution of correlation coefficients Q1
Here is the histogram for distribution of correlation coefficients between answers
to a question “Whether the lesson was pleasant to you?” and following factors:
1. Whether it is accessible (clear) an explanation?
2. Whether arranges you the rate of an explanation?
3. Are you tired at a lesson?
4. Lesson atmosphere: is it comfortable, is it pleasant to you at a lesson?
5. Is the statement filled enough by examples?
6. Objectivity of the marks, which have been put down at a lesson.
7. Are you having some questions to the teacher?
8. Do you still want a lesson on this theme?
9. Have you prepared for this lesson?
10. Are you intending to continue studying at home?
11. Accordance of a lesson to tasks of independent (home) work.
12. Is it interesting to you at a lesson?
13. Have you taken out something useful at a lesson or are sorry about spent time?
The most significant factors had appeared (in decreasing order) 1 (0.91), 4 (0.87),
12 (0.83), 13 (0.75) 5 (0.63), 9 and 11 (0.59). Objectivity of marks is only further
(0.51) and inverse correlation – 0.39 for 7 specifies that for the majority the good
lesson is such after which does not remain questions to the teacher (fig.2).
Here is the histogram for distribution of correlation coefficients between answers
to a Feedback question “Whether the teacher is pleasant to you?” and following fac-
tors:
1. Whether lessons were pleasant to you?
2. Estimation by student of the knowledge received at lesson.
3. Is it accessible (clear) an explanation?
Econometric Analysis on the Site “Lesson Pulse” 377
1
0,8
0,6
0,4
0,2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
-0,2
-0,4
Fig. 2. Histogram for distribution of correlation coefficients Q2
4. How much are accessible (clear) and authentic answers to questions of students?
5. The explanation is filled enough by examples.
6. Using of various approaches at training.
7. Whether the teacher aspires to interest, motivate students?
8. Lessons atmosphere: is it comfortable, is it pleasant to you at a lesson?
9. Availability of the teacher, his inclination to listen to students, to conduct discus-
sions with them.
10. Knowledge of a subject by the teacher.
11. Insistence (regular and frequent enough control of knowledge).
12. Punctuality (comes in time at lessons).
13. Possession of an audience (students do not sleep and do not make too much noise
at lessons).
14. Objectivity in estimation of the student by the teacher. Whether criteria of esti-
mation in all subgroups are identical?
15. Accordance of a lesson to control tasks.
The most significant factors appear (in decreasing order) 8 (0.92), 7 (0.85), 6
(0.775), 4 (0.75), 9 (0.72), 3 (0.675), 13 (0.58).
Only further with factor of correlation 0.51 follows 1 - whether lessons were pleas-
ant to you. And major factors of estimations of the teacher and lesson are considera-
bly differing. Further the histogram of differences between factors of correlation for
questions “Whether the teacher is pleasant to you?” and “Whether lessons were pleas-
ant to you?” is resulted.
The factors much more essential at an estimation of a teacher, than a lesson are 6
(using of various approaches at training) and 7 (whether the teacher aspires to interest,
motivate students). On the contrary, at an estimation of a lesson it is much more es-
sential factors 14 (accordance of a lesson to control tasks) and 10 – insistence (regular
and frequent enough control of knowledge): probably, according to students, insis-
tence it is good for lesson and it is not so good for the teacher.
378 A. J. Weissblut
0,6
0,4
0,2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
-0,2
-0,4
-0,6
Fig. 3. Histogram for distribution of correlation coefficients Q3
Certainly, the correlation matrix contains decomposition on factors also for each of
15 questions. So it is found out that 5 (the explanation is filled enough by examples)
is most closely connected with 15 (accordance of a lesson to control tasks); 3 (are you
tired at a lesson) with 7 (presence of questions to the teacher); 13 (possession of an
audience) with 14 (objectivity in estimation of the student).
It is interesting to compare 12 (is it interesting to you at a lesson) with 13 (have
you taken out something useful at a lesson) from interrogation about results of the
lesson (fig.4).
1
0,8
0,6
0,4
Interesting
0,2 Useful
0
1 2 3 4 5 6 7 8 9 10 11 12
-0,2
-0,4
Fig. 4. Histogram for distribution of correlation coefficients Q4
As we see, from the student’s point of view, it is interesting and it is useful is not
the same. So 4 (lesson atmosphere) correlates with the factor interesting much more,
while factor 5 (is the statement filled enough by examples) with 11 (accordance of a
lesson to tasks of independent (home) work).
Econometric Analysis on the Site “Lesson Pulse” 379
3 The Analysis of Interrogations about the Factors Influencing
a Lesson
Unlike interrogations about results of lesson and Feedback results of interrogations
about factors of influence on a lesson course are close enough in different groups. The
histogram for distribution of interrogation requisites on the relation to lesson (in 421
group) is below (fig. 5).
10
9
8
7
6
5
4
3
2
1
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Fig. 5. Histogram for distribution of correlation coefficients Q5
Here:
1. Is study pleasant to you? Is it interesting to you?
2. Whether you believe that education is “road to the future”?
3. Is your speciality pleasant to you?
4. Is the program of training for your speciality satisfying you?
5. Whether satisfies you teaching level at the university?
6. Whether on own will you have chosen university and a speciality?
7. Would you like to change the speciality or to receive additional higher education?
8. Whether your attendance of lessons is regular?
9. Do you regularly prepare homework?
10. Whether there were at you conflicts to teachers?
11. Were you afraid of an exception of university?
12. Do you wish to take part in scientific work, in Olympiads on your speciality?
13. Whether often to you fellow students address for the help in lessons?
14. Do you wish to enter postgraduate study after training end?
15. What’s the time you spend for preparation for lessons (on the average hours per
day)?
And further similar results of interrogation on external factors (fig. 6):
380 A. J. Weissblut
10
9
8
7
6
5
4
3
2
1
0
1 2 3 4 5 6 7 8 9 10
Fig. 6. Histogram for distribution of correlation coefficients Q6
Here:
1. Close dialogue with teachers.
2. Accessibility of the Internet at university.
3. Readiness of an auditorium for a lesson (working capacity of projectors, com-
puters, the software; comfort of an auditorium).
4. Presence of enough points for the centralized feeding.
5. Accessibility of contacts to the future employers.
6. Accessibility of summer improvement.
7. Participation in scientific work.
8. Teaching level at the university.
In a correlation matrix under all these factors there are only few factors which cor-
relations are close to 1. These are factors:
1. Whether your attendance of lessons is regular with factors
(a) do you regularly prepare homework (0.87)
(b) teaching level at the university (0.84)
(c) participation in scientific work (0.63)
(d) have you prepared for this lesson (0.59)
(e) accessibility of summer improvement (– 0.5)
2. Do you regularly prepare homework with factors
(a) whether your attendance of lessons is regular (0.87)
(b) teaching level at the university (0.815)
(c) have you prepared for this lesson (0.66)
(d) participation in scientific work (0.56)
(e) accessibility of summer improvement (– 0.52)
Econometric Analysis on the Site “Lesson Pulse” 381
3. Teaching level at the university with factors
(a) whether your attendance of lessons is regular (0.843)
(b) do you regularly prepare homework (0.815)
(c) whether on own will you have chosen university and a speciality (0.65)
(d) have you prepared for this lesson (0.59)
(e) participation in scientific work (0.56)
(f) accessibility of summer improvement (– 0.55)
Besides them correlation factors above 0.7 appear still only twice: between factors
whether there were at you conflicts to teachers and were you afraid of an exception
of university (0.85); and between factors participation in scientific work and is
the program of training for your speciality satisfying you (0.74). Occurrence in
such line the factor teaching level at the university is, probably, the best compli-
ment for Faculty of physics, mathematics and informatics of the Kherson state univer-
sity for all its history. Our main task is to use the mental orientation, fixed thus in the
correlation analysis of factors, for separating true students, for which educational
process is a considerable part of their life, from those, who would prefer to keep far
away from it. Using already cited data and the following table 1
Table 1. Correlation analysis of factors
Factor Average value Root-mean-square deviations
Teaching level at the university 702 2.17
Regularly attendance of lessons 8.85 2.3
Regularly prepare homework 8.4 2.6
We choose as a differentiating sign between groups the factor regularly prepare
homework. In this case mutual correlations of defining sign are closer to 1; and the
dispersion is more, that testifies about more variability of respondents under this fac-
tor. Besides, among others selected it more corresponds to such sign on common
sense.
4 Results of Interrogations about Lesson and Feedback on
Subgroups
To the selected differentiating sign among 20 respondents of group 421 the 12 par-
ticipants is allocated, who for a question do you regularly prepare homework have
answered with 10 or 9 points. The additional subgroup consists of 8 respondents.
Whether there correspond such subgroups to required division into true students and
the others? Below there is the histogram for average results of interrogation about
lesson on the allocated subgroups (fig. 7).
382 A. J. Weissblut
12
10
8
Prepare
6
Don't prepare
4
2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Fig. 7. Histogram for distribution of correlation coefficients Q7
So, the factors considerably different in subgroups (in decreasing order of modules
of differences between average values in subgroups) are:
2 Whether it is accessible (clear) an explanation? (7.92 – 4.25 = 3.67)
1 Whether the lesson was pleasant to you? (7.1 – 4.38 = 2.72)
12 Accordance of a lesson to tasks of independent (home) work. (9.91–7.5 = 2.41)
6 Is the statement filled enough by examples? (9.41 – 7 = 2.41)
10 Have you prepared for this lesson? (8.75 – 6.5 = 2.25)
13 Is it interesting to you at a lesson? (7.92 – 5.87 = 2.05)
14 Have you taken out something useful at a lesson? (9.1 – 7.4 = 1.7)
5 Lesson atmosphere (7.66 – 1.25 = 1.41)
9 Do you still want a lesson on this theme? (3.66 – 2.65 = 1.01)
The averages of additional group are more only twice, there are:
4 Are you tired at a lesson? (6.5 – 7.62 = – 1.12)
3 Whether arranges you the rate of an explanation? (5.5 – 6.37 = – 0.87)
Last result is strange at first sight, but steady for all groups and it is easy to explain
this phenomenon psychologically: as less adjusted the student for study, the more he
would like acceleration, faster course of time.
Further there are similar results for interrogation Feedback (fig. 8).
12
10
8
Prepare
6
Don't prepare
4
2
0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
Fig. 8. Histogram for distribution of correlation coefficients Q8
Econometric Analysis on the Site “Lesson Pulse” 383
Here the factors considerably different in subgroups are:
5 The explanation is filled enough by examples (9.17 – 6.37 = 2.8)
6 Using of various approaches at training (6.36 – 4 = 2.36)
16 Accordance of a lesson to control tasks (8.9 – 6.87 = 2.03)
4 How much are accessible (clear) and authentic answers? (7 – 5.25 = 1.75)
9 Lesson atmosphere (6.36 – 4.85 = 1.51)
The obtained data corresponds to a hypothesis about required division into groups,
anyway they don’t contradict it.
5 The Latent Division in Group
The site “Lesson pulse” offers also division of group into classes with a given value
of mutual correlation: between two respondents from one class it is possible to find a
chain of respondents of this class so, that the correlations of answers between con-
secutive respondents of this chain not less than the given value. Such division into
subgroups allows finding out distinctions in the group, which is not appreciable di-
rectly.
At mental interrogation about factors of influence on lesson and the set minimum
level of mutual correlation 0,6 in test group 421 splitting into 3 classes has turned out:
from 4, from 5 and from basic subgroup of 11 respondents. Let's compare averages of
the basic class to averages of the first and second subgroups under those factors in
which appreciable differences have come to light.
1. Is the program of training for your speciality satisfying you?
2. Would you like to change the speciality or to receive additional higher education?
3. Do you wish to take part in scientific work, in Olympiads on your speciality?
4. Do you wish to enter postgraduate study after training end?
5. Participation in scientific work.
6. Readiness of an auditorium for a lesson.
7. Accessibility of summer improvement.
8. Accessibility of contacts to the future employers.
12
10
8
Class 1
6 Class 2
Main class
4
2
0
1 2 3 4 5 6 7 8
Fig. 9. Histogram for distribution of correlation coefficients Q9
384 A. J. Weissblut
Respondents from classes 1 and 2 much less than the basic group are satisfied by
the program of training of the speciality (point 1). They would like to change the spe-
ciality or to receive additional higher education essentially more than the basic group
(point 2). Their difference is the sharpest comes to light in point 3: unlike the basic
group they at all do not wish to take part in scientific work or in the Olympiads on the
speciality. So, apparently, the speciality for them has lost now appeal. Respondents
from a class 2 does not interest in the postgraduate study (point 4), however they do
not against scientific work (point 5). The main thing, they have the most interest in
contacts to employers (point 8). Apparently, it is search of the application out of the
speciality. Respondents from a class 1 are focused differently: they have a little inter-
est in scientific work and employers (points 5 and 8), but they wish to enter post-
graduate study (point 4).
References
1. De Bono, E.: Six Thinking Hats. Penguins Books (1997)
2. PHP Book, http://www.phpreferencebook.com (2012)
3. Hansen, B. E.: Econometrics. Textbook, http://www.ssc.wisc.edu (2012)
Decision Supporting Procedure for Strategic Planning:
DEA Implementation for Regional Economy Efficiency
Estimation
Karine Mesropyan1
1
Institute of Socio-Economic and Humanitarian Researches of Southern
Scientific Center of RAS, 41 Chekhova prospect, 344006 Rostov-on-Don, Russia
karineemesropyan@gmail.com
Abstract. The algorithm of decision supporting procedure based on Data En-
velopment Analysis (DEA) along with the Malmquist Productivity Index is
suggested in the paper. The procedure’s core consists of evaluations complex
for preliminary data processing and adjustment as well as creating of analytic
materials in the field of regional strategy planning. The crucial study issue is to
define boundaries of DEA applicability in this field and to eliminate DEA
shortcoming, such as the scores dependence on a set of inputs and outputs. The
efficiency scores of Russian regional agrarian sector are obtained in order to
verify the procedure and add knowledge to current indicators’ systems of re-
gional economic efficiency by improving approach objectiveness. It is shown
how obtained results can be applied in the strategic planning to increase effec-
tiveness of state regional policy activity.
Keywords. Efficiency, Malmquist Index, Data Envelopment Analysis, Region,
Procedure, Strategy Planning
Key terms. DecisionMaking, MathematicalModel, Methodology, Develop-
ment, Management
1 Introduction
Theoretical model of this study is based on the Pareto-Koopmans concept (1951) [1].
System technology is efficient by Pareto-Koopmans if and only if the object does not
have an opportunity to improve its resource (input) or product (output) without sacri-
ficing some other input or output. Charnes et al. (1978) have proposed Data Envel-
opment Analysis (DEA) based on this concept of efficiency that was combined opera-
tional research tools within works of Koopmans (1951) and Farrell (1957) [2, 3].
DEA is a non-parametric frontier approach for comparative efficiency measure-
ment in which a set of similar objects with multiple inputs and outputs is analyzed.
386 K. Mesropyan
The aim of this study is to suggest the procedure for providing strategy planning by
analytical reports based on DEA scores implementation.
It is obvious that productivity analysis by DEA has at least three current issues.
The first is to define a set of objects which will be compared in the study. The second
is to formulate convenient conditions for concrete models’ modifications using. The
third issue is to improve discrimination capability. Therefore, efficiency assessment
procedure by DEA is primarily based on the following grounds: the formation of ob-
jects’ set to be compared, identification of inputs and outputs, and model selection.
Taking into account issues mentioned above, it is necessary to adapt basic DEA
models and its implementation. Furthermore, DEA procedure is considered as a core
of the evaluation of regional economy efficiency scores.
This paper consists of five parts. We state the main issues in this, first, part. This
study’s background is presented in the second part. Part 3 deals to description of sug-
gested evaluation procedure. The applying of investigation of the Russian regions
agrarian sector to management tasks by using the procedure is reported in the part 4.
We make conclusions in the last part.
2 Theoretical and Methodological Background
DEA application has a big number of advantages. First of all, a calculation of an inte-
grated assessment is produced for each region reflecting the efficiency of input factors
using for output products. Besides, the Pareto-optimal set of efficient regions in the
multidimensional space of inputs and outputs is being obtained. Secondly, it is unnec-
essary to attract an expert knowledge in a priori assignment of weights for variables
corresponding to inputs and outputs. Despite of this, using of additional data on re-
gion external factors is helpful for creating the right model. Thirdly, it is very impor-
tant that there are no restrictions on the functional form of the relation between inputs
and outputs.
The study is carrying out with the hypothesis that DEA implementation needs the
formal procedure in order to obtain stable scores and apply research results to analytic
background of current regional strategic planning.
The multilateral and penetrating analysis of DEA possibilities and its application’s
restrictions are presented in Dyson et al. (2001), Cook and Seiford (2009) [4, 5].
Along with these works there are reviews of this method application, for instance, in
papers of Avkiran and Parker (2010), Liu et al. (2012) [6, 7]. The common bases pro-
ductivity measurement presented in Caves et al. (1982) [8].
The application possibilities of Malmquist Productivity Index in different in-
tertemporal comparisons are described in the research of Färe and Grosskopf (1996)
[9]. Tsuneyoshi et al. (2012) used Malmquist Index for the comparative analysis of 97
countries calculated by DEA models for period 1981-2004 [10]. Yamamura and Shin
(2008) determined the nature of inequality impact on capital accumulation and growth
performance by evaluation DEA indexes from 1965 to 1990 [11].
According to review presented in [12], although there are a DEA advantages, the
general method’s shortcoming is considered as crucial because the scores signifi-
Decision Supporting Procedure for Strategic Planning … 387
cantly depend on a set of inputs and outputs. This study suggests the special proce-
dure for DEA implementation for the needs of regional strategic planning. It is a re-
sult of attempting to eliminate the mentioned DEA drawback and provide the decision
process of strategic planning by analytic materials. Golany and Roll (1989), Em-
rouznejad and De Witte (2010) offered procedures of DEA application which are very
useful for common case [13, 14]. This study based on results of these works.
3 Efficiency Estimation Procedure
Different levels of the regional economy scale and the return to scale effect are con-
sidered as a reason of inequality between regional output performances. That is why
the model with variable return to scale is suggested for this study. This model was
introduced by Banker et al. (1984).
Data for a research by DEA is presented by a number of indicators in form of the
matrix of inputs Xt={xtij} and matrix of outputs Yt={ytkj}. The efficiency criterion for a
multidimensional assessment of an object is to assign some input and output parame-
ters for all objects, some weights and then to calculate and maximize the ratio for
each object:
(1)
where:
j – index of the estimated production facility, j=1…, n;
xij – matrix of input parameters that reflect the system resources, i=1,…,m;
ykj – matrix of outputs which reflect the products of system, k=1,…,s;
ui /wk – weights for outputs/inputs.
According to the DEA framework, this function should be maximized under re-
strictions for all objects:
(2)
The Malmquist Productivity Index is calculated using such DEA efficiency scores
for evaluation of total factor productivity change:
(3)
The suggested procedure for regional efficiency assessment has the complex of
procedures for preliminary data processing and adjustment (fig.1).
388 K. Mesropyan
Fig. 1. Five stages of Efficiency Estimation Procedure
The dual linear program model for evaluation the criterion given above is:
(4)
where:
η – comparative efficiency score of region j (j=1…, n);
λj – dual model variables.
If η<1 then the region belongs to inefficient set, otherwise (η = 1) it is a part of
Pareto set.
According to procedure carrying out, the result of all evaluations using (3), (4) is a
set of regional types by dynamics character.
Generally, the result of DEA application is the set of scores also which shows the
ways of comparative efficiency improvement for each inefficient region. Neverthe-
less, this issue is not treated in this procedure because it requires the special attention
and investigation due to its complexity.
4 Evaluation by Using Procedure
We examined the issue of inequality of regional economy performance for the period
from 2008 to 2010. Federal State Statistics Data is used from www.gks.ru [15]. We
Decision Supporting Procedure for Strategic Planning … 389
evolved a set of indicators that can be used in a broader context in order to identify
factors influencing on a regional underdevelopment.
Stage 1. The set of indicators for models consists of resources and results of re-
gional agrarian sector performance. Indicators are reported in Table 1.
Stage 2. Set of model’s variables defined this way: five resources are taken as in-
puts while three results are taken as outputs. The volume of region population is con-
sidered as the special variable for the set’s normalization.
Table 1. The set of regional performance indicators
Type Indicator
Resources number of cattle, thousand heads (x1)
organizations acreage under crops, ha (x2)
average number of employees, thousand people (x3)
power capacity, thousand horsepower (x4)
equipment park (tractors), units (x5)
Results gross grain yield, thousand tons (y1)
production of milk, thousand tons (y2)
production of livestock and poultry, thousand tons (y3)
Variable for Nor-
malization volume of region population, thousand people
Stage 3. Next, the homogeneity of conditions was checked for all agrarian regional
systems, and asymmetry of land’s quality founded out. The input called “organiza-
tions acreage under crops” is adjusted by the coefficient of cadastral value of agricul-
tural land. The rule of ratio of variables’ number and objects’ number is kept, there-
fore, modeling is made for 53 quite similar agrarian Russian regions.
Additional restriction to weights is used in order to improve the discrimination ca-
pability of the model. Direct restriction on the ratio of the quantity of employees and
the power capacity presented by the following ratios:
(5)
(6)
(7)
where:
390 K. Mesropyan
price_x3 – regions’ average monthly salary;
price_x4 – regions’ average energy price;
x3, x4 – inputs which are taken in the normalized forms according to the second
stage.
Stage 4. According to (3) and (4), the calculation cycle is done, and obtained
scores are insensitive regarding the model parameters changes. It was approved by
decreasing of the set of analyzed objects. Besides, the Malmquist Indices values are
similar to current expert opinion on the character of current tendencies of technologi-
cal progress changes in the industry for analyzed period.
Stage 5. The quantitative scores combined with qualitative evaluation of risks and
conditions of regional development allow finding out the regions taxonomy by using
obtained knowledge on type of efficiency dynamics. The procedure has conducted
from the first to the fifth stage given in Fig.1.
The most significant agrarian regions of Russia are located on the Southern terri-
tory which consists of two state districts, namely Southern Federal District and North-
ern Caucasus Federal District. There are two strategies for these regions: Southern
Federal District Strategy and Northern Caucasus Federal District Strategy [16, 17].
Obtained results can be part of analytic reports of these policy development docu-
ments (tables 2-3). Development scores are presented for period 2008 - 2010 in the
tables. The indicator is equal to «+»/«–» in the case of positive/negative dynamics.
Table 2. Efficiency Scores for the Southern Federal District Strategy
Region Development Region
Score Type
Republic of Adygeya ++ Stable Growth
Republic of Kalmykiya ++ Stable Growth
Krasnodar Region –+ Unstable Decline
Astrakhan Region +– Unstable Growth
Volgograd Region ++ Stable Growth
Rostov Region +– Unstable Growth
Table 3. Efficiency Scores for the Northern Caucasus Federal District Strategy
Region Development Region
Score Type
Daghestan –– Stable Decline
The Ingush Republic ++ Stable Growth
Republic of Kabardino-Balkariya –– Stable Decline
Republic of Karachaevo-Cherkesiya –– Stable Decline
Republic of Northern Osetia Alaniya –+ Unstable Decline
The Chechen Republic –– Stable Decline
Stavropol Region ++ Stable Growth
Decision Supporting Procedure for Strategic Planning … 391
Although analyzed period covers the crisis years, the agrarian production of the
South of Russia shows the reserve of stability. Besides, it is brought out that the
Southern regions belong to Pareto-Efficient set of Russian regions.
Thus, only 4 regions among 13 of the South are estimated as having the stable de-
cline. The economic development opportunities of this regions are significant, never-
theless the considerable potential of regions is not using.
Such indicators’ further analysis can be used for adjustment of scenario data tasks
in the field of regional development foresight and strategic planning. The obtained
results also can be suitable for equalization policy design in order to steady the level
of regional efficiency during long-term period. In addition to this, it is important that
risks, conditions and possible consequences of the policy should be assumed for each
scenario of regional development.
5 Conclusions
The verification of suggested procedure along with the DEA model demonstrates the
positive results that approve the possibility of the procedure application to prospective
studies in the field of production analysis as well as strategic management.
As it was shown, the obtained results can be part of the quantitative investigations
for the current strategies of development policy. In addition to this, the development
scores can be used together with regional development risks and opportunities analy-
sis, indicators of economical efficiency, such as gross domestic product per capita,
enterprises profitability, etc. Thus, obtained scores will add knowledge to current
indicators’ systems of regional economic efficiency and improve approach objective-
ness and effectiveness of state regional policy activity.
This article is conducted within Program of the Presidium of Russian Academy of
Sciences № 32 “Fundamental Issues of Polyethnic Region Modernization in Terms of
Tensions Growth”.
References
1. Koopmans, T. C.: An Analysis of Production as an Efficient Combination of Activities.
Activity Analysis of Production and Allocation. Cowless Comission for Research in Eco-
nomics. Monograph No. 13, New York: Wiley, pp. 15–32 (1951)
2. Farrell, M. J.: The Measurement of Productive Efficiency. J. of the Royal Statistical Soci-
ety, Series A (General), Part III. 120, 253–281 (1957)
3. Charnes, A., Cooper, W. W., Rhodes, E.: Measuring the Efficiency of Decision Making
Units. European J. of Operational Research, 2, 429-444 (1978)
4. Cook, W. D., Seiford, L. M.: Data Envelopment Analysis (DEA) – Thirty Years On. Euro-
pean J. of Operational Research, 192, 1–17 (2009)
5. Dyson, R. G., Allen, R., Camanho A. S., Podinovski V. V., Sarrico C. S., Shale E. A.: Pit-
falls and Protocols in DEA. European J. of Operational Research, 132, 245–259 (2001)
6. Avkiran, N. K., Parker, B. R.: Pushing the DEA Research Envelope. Socio-Economic
Planning Sciences, 44, 1–7 (2010)
392 K. Mesropyan
7. Liu, J. S., Lu, L. Y. Y., Lu, W.-M., Lin, B. J .Y.: Data Envelopment Analysis 1978–2010:
A Citation-Based Literature Survey. Omega (2012)
8. Caves, D. W., Christensen, L.R., Diewert, W.E.: The Economic Theory of Index Numbers
and the Measurement of Inputs, Outputs and Productivity. Econometrica, 50(6), 1393–
1414 (1982)
9. Färe, R., Grosskopf, S.: Intertemporal Production Frontiers: With Dynamic DEA. Kluwer
Academic, Boston, MA (1996)
10. Tsuneyoshi, T., Hashimoto, A., Haneda, S.: Quantitative Evaluation of Nation Stability. J.
of Policy Modeling, 34, 132–154 (2012)
11. Yamamura, E., Shin, I.: Effects of Income Inequality on Growth through Efficiency Im-
provement and Capital Accumulation. MPRA Paper No. 10220, http://mpra.ub.uni-
muenchen.de/10220/ (2008)
12. Mesropyan, K., Goryushina, E.: Economic Heterogeneity and Political Instability: Experi-
ence and Prospects for Cross-Country Comparisons. The Region Economy: Problems,
Findings, Prospects, Issue 13. OON RAN, ISERH SSC RAS, Volgograd, 58–66 (2012) (in
Russian)
13. Emrouznejad, A., De Witte, K.: COOPER-Framework: A Unified Process for Non-
Parametric Projects. European J. of Operational Research, 207(3), 1573–1586 (2010)
14. Golany, B., Roll, Y.: An Application Procedure for DEA. Omega, 1(3), 237–250 (1989)
15. Federal State Statistics Data, www.gks.ru (in Russian)
16. Southern Federal District Strategy, http://www.minregion.ru (in Russian)
17. Northern Caucasus Federal District Strategy, http://www.minregion.ru (in Russian)
Applying of Fuzzy Logic Modeling for the Assessment
of ERP Projects Efficiency
Andriy Semenyuk
Lviv Academy of Commerce, Tugan-Baranovskoogo. 10, 79005 Lviv, Ukraine
andriy.semenyuk@gmail.com
Abstract. ERP software is one of costly and crucial projects for business in-
vestment. It is known that nowadays Enterprises can rarely afford to implement
long-term projects, in most cases the duration of implementation varies from 3-
4 months (automation of individual store retail chain) and 1-1.5 years when it
comes to big projects. Only successful combination of analytical tools and
methodologies will allow the project to realize and implement ERP-solutions
for commercial enterprises on time and according to set business requirements.
This paper proposes a practical assessment model which applies both the fuzzy
analytic logic approach and the Expert Judgment method to evaluate whether
the ERP software implementation project has succeeded or not.
Keywords. Modeling, Efficiency, Project, Information Technology, Implemen-
tation, Enterprise, ERP
Key terms. Model, Methodology, Process, Management, Approach
1 Introduction
Today it is especially important for Ukrainian Enterprises to be capable to analyze all
the environmental aspects within they operate and be able to plan all needed resources
with the most possible accuracy. To gain such competitive advantages within chang-
ing economy and unstable markets, for particular Enterprise is not enough just to have
the most modern production lines or good educated personnel it also requires to pos-
sess some advanced technologies and modern information management systems that
will quickly allow to react and adapt all the further changes. That is why more and
more Enterprises in Ukraine from different sectors of economy are choosing to im-
plement the Enterprise Resource Planning (ERP) system.
ERP plays an important role to integrate organization's information and functions
in all areas of enterprise activities and within the variety of departments and finally
results in successful operation on the market. However ERP implementation as a pro-
ject itself is costly and time consuming, it also can lead to loss of many valuable re-
sources of the company in case of wrong approached and not efficient way of imple-
394 A. Semenyuk
mentation. So it is critically important for the Enterprises to understand and clearly
realize all the value achieved from ERP initiative [2].
Many factors are essential in determining the efficiency of the ERP projects. Since
most of these factors are qualitative and relations between them are very complicated,
determining their exact quantitative values is quiet difficult. Using the combined
methods of Expert Judgment and Fuzzy logic can be helpful to simplify the calcula-
tions and finally leads to a more precise result to determine generalized efficiency
value of the implemented ERP-project. In this research, we intended to gather optimal
KPIs values from different Enterprise activities and departments based on Experts
Judgment data and design a combined Fuzzy Model to assess the efficiency for ERP-
project [4].
2 Problem Statement and Goals of the Paper
The most common aspects and issues related to ERP systems development and im-
plementation methods of ERP-projects seems to be widely discovered observed and
investigated in works of many foreign as well as Ukrainian researches and scientists
[1], [2], [4], [5]. However the problems related with particular to the efficiency of
such projects are not enough covered. In view of this subject of the research is still
topical.
The main aim of the paper is to present a model for evaluating the effectiveness of
ERP-project implemented on the markets of underdeveloped economic systems with
involves a combination of fuzzy logic and expert judgment methods. Because the
initial data for measuring the effectiveness of ERP-project is mostly inaccurate and
variable so the use of fuzzy logic techniques to enhance the data gathered from the
expert is really feasible here.
3 Proposed Model and Approach
In order to develop the model for the assessment of ERP-project efficiency first of all
it was conducted a set of related Expert Judgment questionnaires sessions, for the
proposed assessment methods and optimal performance indicators or key performance
indicators (KPI) values, in other words KPI is a type of performance measurement.
An organization may use KPIs to evaluate its success, or to evaluate the success of a
particular activity in which it is engaged. Sometimes success is defined in terms of
making progress toward strategic goals, but often success is simply the repeated, peri-
odic achievement of some level of operational goal (e.g. zero defects, 10/10 customer
satisfaction, etc. Accordingly, choosing the right KPIs relies upon a good understand-
ing of what is important to the organization [9]. The basics input data of our research
are KPIs of ERP-projects that were received from managers of ERP-projects, and
classified according to the criterion of "trend change".
Determination of further scope of KPIs and the major ERP success factors also is
based on variety international consulting agencies reports KPIs values, and ERP-projects
Applying of Fuzzy Logic Modeling for the Assessment … 395
statistics obtained from such reports was additionally verified with the involved experts
[6], [7], [8].
As the result it has been allocated four main groups of indicators: 1) X - perform-
ance "increase group of KPIs" (the actual value of which increased for the Enterprise
after ERP implementation), 2) Y - values "reduce group of KPIs" (the value of which
decreased for the Enterprise after ERP implementation) 3) W – project financial and
investments indicators. 4) Z - Generally optimized qualitative KPIs for different proc-
ess aspects within Enterprise.
To be able assess the effectiveness of ERP-project we developed a structural com-
bined model (depicted on Fig. 1). The model contains of methodological approaches
and structure of key performance indicators to determine the effectiveness of ERP-
project. Summarized decision tree inference, which is presented at the bottom of the
model, reflects the hierarchy of input variables.
Fig. 1. Combined Model of ERP Projects Efficiency Assessment
Developing a questionnaire for determining the Fuzzy (if-then) Rules with respect
to the three optimal KPIs values obtained from Experts as inputs and ERP project
efficient value as output. Validity and reliability of the questionnaire was confirmed
and they were distributed among the ERP practitioners. Fuzzy rules as the basis for
determination of the conditions of the company KPIs have been formed and entered
into Fuzzy system through MATLAB software.
Currently there are variety of software tools and system for applying the fuzzy
logic calculations available on the market (CubiCalc 2.0 RTC, CubiQuick, FIDE, Flex
Tool, FuziCalc, FuzzyTECH, JFS, MATLAB - Fuzzy Logic Toolbox, RuleMaker
etc). Each of these products has its own strengths and weaknesses, however as soft-
ware platform for our research it was decided to go with MATLAB Fuzzy Logic
396 A. Semenyuk
Toolbox (FLT) as most appropriate tool in particular because of the integrated nature
of the MATLAB environment that also provides functions, applications, and a simula-
tive block for analyzing, designing, and simulating systems based on fuzzy logic,
widely used not only in academic and research institutions but by industrial enter-
prises as well.
To determine the value of each factor, different questionnaire was prepared to collect
related information for the KPIs values. Also validity and reliability of mentioned
questionnaire was confirmed and it was distributed among managers and experts in that
domain. Calculated means obtained from questionnaire, results has been inputted to the
Fuzzy System (see fig. 2). Final results were analyzed and the efficiency of the imple-
mented ERP project was determined by this software.
Fig. 2. Identifying the membership functions for each input and output variable
To design a Fuzzy system by MATLAB following: X, Y, Z, W are lingual values that
also represents an organizational major KPIs groups and E is the lingual value of summa-
rized project Efficiency (see fig. 3 and fig. 4).
Fig. 3. The Fuzzy system of assessing the efficiency of the ERP implementation in MATLAB
software
Applying of Fuzzy Logic Modeling for the Assessment … 397
Fig. 4. Drawing a conclusion in Fuzzy system
4 Conclusions
Proposed approach and model of determining the efficiency level of the implemented
ERP-project with fuzzy logic can be used by variety of ERP practitioners, project
managers, top management personnel of Enterprises that implements ERP-system for
advanced analysis of the actual project results. Fuzzy rules should be further validated
and formed by consulting with ERP practitioners and their results will be entered into
knowledge base of Fuzzy system. Methods of fuzzy logic computing combined with
provided expert judgment aimed optimize the speed of the decision-making on project
efficiency level and simultaneously provide more accurate assessment abilities.
References
1. Finney. S., Corbett. M.: ERP Implementation: a Compilation and Analysis of Critical
Success Factors. Business Process Management Journal, 13(3), 329–347 (2007).
2. Allen, D., Ken, T., Havenhand, M.: ERP. Critical Success Factors: an Exploration of the
Contextual Factors in Public Sector Institution. In: Proc. 35th Hawaii International Confer-
encr on System Sciences, pp. 244–247 (2002)
3. O'Leary, D.: ERP Systems: Modern Planning and Enterprise Resource Management. Se-
lect, Implement, Utilize. Vershina, Moscow (2004)
4. Savavko, M.: IS Fuzzy Expert. Publishing House of I. Franko Lviv National University,
Lviv (2007).
5. Nozdrina, L.: Applying of Fuzzy Logic Modeling for the Assessment of ERP Projects Ef-
ficiency. In: Proc. 5th Int. Sci. Conf. Project Management: Status and Opportunities, pp.
1–2, NUS, Nikolaev (2009)
398 A. Semenyuk
6. Gartner: Information Technology Research and Advisory Agency.
http://www.gartner.com/technology/home.jsp
7. Panorama Consulting. Consulting Firm with Expertise in ERP Software, http://panorama-
consulting.com
8. IDC. Global Provider of Market Intelligence, Advisory Services, and Events for the Infor-
mation Technology, http://www.idc.com/home.jsp?t=1365517508962#.UWQk-5NTCjg
9. Austin, R. D.: Measuring and Managing Performance in Organizations. Dorset House Pub-
lishing (1996)
Applying of Fuzzy Logic Modeling for the Assessment … 399
Appendix A. Example of ERP User Satisfaction Survey
400 A. Semenyuk
Appendix B. Example of ERP Expert Judgment Questionnaire
Mathematical Model of Banking Firm as Tool
for Analysis, Management and Learning
Victor Selyutin1,2 and Margarita Rudenko1
1
Research Institute of Mechanics and Applied Mathematics,
vvs1812@gmail.com, ritusik@mail.ru
2
Economic Faculty of Southern Federal University
220/1, av. Stachki, 344090, Rostov-on-Don, Russia
Abstract. An essential concern for banking firms is the problem of assets and
liabilities managing (ALM). Over last years a lot of model tools were offered
for solving this problem. We offer the novel approach to ALM based on trans-
port equations for loan and deposit dynamics. Given the bank's initial state, and
various deposit inflow scenarios the model allows provide simulations includ-
ing stress-testing, and can be used for assessment of liquidity risk, for examine
loan issue decisions to choose reasonable solution, and in the learning purposes.
Keywords. Asset- liability management, Differential equations, Liquidity risk,
Duration
Key terms. Banking, Mathematical Modelling, Decision making
1 Introduction
A banking firm is rather a complex system within the context of management prob-
lem. It is caused by a considerable number of financial flows and the funds, having a
various origin and differing by dynamic and probabilistic characteristics, and at the
same time forming the unified system. Stable functioning of the system is provided
due to hierarchy, external (prudential supervision) and internal regulators and restric-
tions, and feedbacks.
Among the mathematical models of banking firms it is possible to separate two ba-
sic groups. There are models of optimization of assets portfolio (static, single and
multi-period) using linear and dynamic programming mainly [1-2], and models of
assets and liability management (ALM), using methodology and the technique of the
stochastic differential equations [3-5].
One of the problem solved by models ALM is management of various risks (espe-
cially credit risk and interest-rate risk), including the problem of default probability
decrease.
402 V. Selyutin and M. Rudenko
In connection with computer engineering development, from the middle of 70th
years of the last century the computer models of banks focused on problems of plan-
ning and decision-making support systems began to appear. However such projects
had no further development [6-8].
Then we will turn our attention to one of the possible approaches to bank model-
ling as a dynamic system, which can be called hybrid. The main tasks which the
model developed must solve are the analysis and management of liquidity and stress-
testing of a bank. In addition, it can be used for optimization of assets profile.
Aggregation of elements of balance sheet can be varied according to the objectives
of modelling and principles developing of state variables vector. We will use the fol-
lowing simplified schematic (Tab. 1).
Fixed assets of bank we will ignore, taking into account only financial flows. Ob-
viously, balance sheet equation takes place:
A= S+B+Q+X=Y+C+M=L , (1)
where equity (capital) of a bank C is a balancing variable.
For detailed modelling of credit risks, loans issued can be divided by categories of
the debtors having various reliabilities. Division of deposits on time and demand is
necessary for calculation of instant liquidity. It is ignored in considered below version
of the model for simplicity.
Table 1. The aggregated balance sheet of commercial bank.
Assets (A) Liabilities (L)
Loans issued Business Debt (Y) Time deposits
(X): Private customers On-demand deposits
(buyer`s credits, mort- and current ac-
gage etc.) counts
Other banks Inter-bank credits (M)
Securities Shares (Q)
Bonds (B)
Reserves (S) Cash Equity, including retained profit
Rest fund, loan loss re- of last periods (C)
serves etc.
Formally it is possible to mark three groups of operations in the balance-sheet ta-
ble:
Reallocation of assets between separate items
Reallocation of liabilities between separate items
Identical change of assets and liabilities at one period
Though the bank opens a position in liabilities with grant of a loan (opening of a
credit line) at one time, from the formal point of view this operation is resolved into
reallocation of asset`s items.
Mathematical Model of Banking Firm … 403
Similarly, if the deposit remains unclaimed in maturity date it either is prolonged,
or is transferred in demand deposits (with no interest accruing or with the minimum
percentage) according to contact conditions. Actually, in this case there is a realloca-
tion of liability`s items.
At last, when interest on loans (or other types of income or expenses) are received
(or repaid), at one time it is changed both assets, and liabilities, own capital of bank
increases or decreases.
2 Model with Certain Terms of Loans and Attracted Funds
The main difficulty in modelling of assets and liabilities dynamics is concerned with
necessity taking into account terms of loans and deposits. Due to these the state vari-
ables must depend on two parameters - current time (t) and current "age" () or the
remained term to maturity (T-). That is why dynamics of the issued loans can be
described by following transport equation:
x x
u ( t , ) (2)
t
T
In addition X ( t ) x ( t , ) d - total amount loans issued,
0
T
X * ( t ) x ( t , ) e d - present value of loans, Т – term of loans.
0
Movement of time deposits is described similarly:
y y
v ( t , )
t (3)
T
Y ( t ) y ( t , ) d - total amount of time deposits,
0
T
Y * ( t ) y ( t , ) e d - present value of time deposits, T – term of deposits.
0
Variables u(t,) and v(t,) denote the flows of issued loans (temporary outflow of
financial resources of bank) and deposits (temporary inflow) distributed by time tak-
ing into account amortization (interest payment or installment credits). Accordingly,
total inputs of loans U(t) and deposits V(t) (or its present values U*(t) and V*(t)) is
described as:
T T
U (t ) u (t , ) d , U * ( t ) u ( t , ) e d
0 0
T T
V (t ) v (t , ) d , V * ( t ) v ( t , ) e d
0 0
Solution of the equations (2-3) can be represented in the closed form:
404 V. Selyutin and M. Rudenko
t
x ( t , ) u ( , t ) d ( t )
0
t
y (t , ) v ( , t ) d ( t )
0
where () and ()- initial distributions of loans and deposits by "age", or may be
obtained by use corresponding equations with finite differences.
Dynamics of reserves (S) and equity (C) is described by the equations including
stochastic members which consider random nature of change in value of shares and
possible loans losses:
dS U t V t X X Y Y B B M M Z t dt Qdt
QdWt xt dJ t
dC X X Y Y B B M M Z t dt Qdt QdW t
xT dJ t
where dWt – increment of Wiener stochastic process, dJt – increment of compound
Poisson process with exponential distributed size of jumps (loan losses), Z(t) – opera-
tion expenses and payment for dividends; xT(t) – repayment of a loans in maturity
date, X, Y, B, M – accordingly interest on loans, deposits, bonds income, cost of
credits; - average portfolio return of trading securities, - volatility of securities
portfolio.
Investments in liquid assets - shares Q(t) and bonds B(t) can be considered as some
parameters of management and to be calculated, proceeding from structure of assets
chosen or planned by bank taking into account loan demand. Similarly, the volume of
received loans M(t) can be select depending on bank`s requirement in financial re-
sources.
It is necessary to add the equations of dynamics of duration to the equations of
movement of assets and liabilities to model liquidity risk taking into account change
of interest rates
If r – the annual interest rate, so in this case Macaulay duration for an asset x(t,) is
defined by expression:
T
1
D x t T x t , e d ,
X * t 0
where = ln(1+r).
Similarly duration of another financial flows y(t,), u(t,), v(t,) are calculated. It is
possible to show that dynamics of duration is described by any of presented below the
equations which is chosen according to liquidity research tasks.
dD x U * (t )
D u (t ) 1 ( t ) D x D x
dt X * ( t )
dD x U * (t ) xT (t )
( D u (t ) D x ) 1 D x D x
dt X * (t ) X * (t )
Mathematical Model of Banking Firm … 405
dD x x (t )
( t )[ D u ( t ) D x ] 1 D u ( t ) T D x
dt X * (t )
where dX * 1
(t )
dt X *
Model (2-3) has been transformed to system of difference equations and realized as
computer program [9]. The user independently chooses one of two operating modes
of the program: calculation in case of predefined planning horizon, or calculation with
possible correction of parameters, setting physical speed of calculation.
The program is interactive as the user can change values of some key parameters in
the process of calculation, without interrupting its work. As key parameters are cho-
sen: a fraction of cash invested in various kinds of assets, revenues (interest rates), a
duration of demand deposits, credit demand, inflow of deposits, crediting scenarios
(distribution of loans by time).
Dynamics of inflows and outflows of cashes; diagram of change of durations of as-
sets and liabilities; distributions of loans and deposits, and also input flow by time are
displayed on the screen of computer.
A stress-testing is provided in the program. The user can choose the period of
stress-testing and such stresses-scenarios as decrease in inflow of deposits, decrease
in duration of deposits (the scenario of outflow of deposits); decrease in accessible
volume of attracted funds on the interbank market.
3 Model with Fixed Terms of Lending and Borrowing
Model (2-3) presented above is rather difficult in numerical realization and does not
allow to consider some important facts, for example, dependence on interest rates
from different terms of lending or borrowing. Therefore we will consider simplified
modification of previous model under supposing that terms of loans and deposits are
fixed.
It is possible to fix the most typical terms according to the classification used in the
bank reporting, in spite of the fact that terms of loans (or deposits) can be arbitrary.
Both loans and time deposits are structured by terms as follows: till 30 days, from 31
till 90 days, from 91 till 180 days, from 181 days till 1 year, from 1 year till 3 years,
over 3 years.
Accordingly, it is possible to establish several typical periods Tk for each of them
time transactions are described by the partial differential equation of the first order
x x
a( , x) (4)
t
with a boundary condition x(t,0)=u(t) and the initial condition x(0,)=(). Initial and
boundary condition should be consistent, that is u(0)=(0).
Here t - current time, 0t<, - elapsed time since the moment of settlement of
transaction ("age" of an loan or deposit), 0<T, a(,x)- value of "amortization" of an
loan or deposit (inflation, installment credit etc.).
406 V. Selyutin and M. Rudenko
Similarly (2-3), the variable x(t,) is the allocated variable characterizing some
credit tools, accounted in assets or in liabilities (loans for limited period, time depos-
its, interbank lending or borrowing, coupon bonds or other assets and liabilities with
the fixed term of repayment).
Further it will be assumed that
a ( , x ) x (5)
i.e. repayment of credits occurs proportionally to their volume with coefficient ,
which is not dependent on age. It can be used and other schemes (when credit repay-
ment begins not at once and (or) occurs in advance established equal shares.
It is easy to verify that the solution of the equation (4) looks like a travelling wave
x(t , ) u (t ) exp(x) (6)
For consistency an initial and boundary conditions at t<T it is necessary to pre-
determine u(t) on an interval t[-T,0).
From (4) - (6) follows
x(0, ) ( ) u( ) exp( ) (7)
and after replacement for -t,
u (t ) (t ) exp(t ) under -Tt<0 (8)
The total value of the considered loan (or deposit) are obtained by integration on
age
T
X ( t ) x (t , ) d
0
(9)
Substituting (6) in (9), we have
T
X ( t ) u ( t ) exp( ) d (10)
0
Integrating (4), we obtain the ordinal differential equation
dX
u (t ) X x (t , T ) u (t ) X u (t T ) exp( T ) (11)
dt
As assets with different terms of repayment are in portfolio of assets or liabilities,
so it is possible to replace scalar variable X(t) in (11) with vector. Vector`s compo-
nents are financial tools with different terms of repayment
dXk
uk (t ) k X k xk (t, Tk ) uk (t ) k X k uk (t Tk ) exp( kTk ) (12)
dt
For simplicity further we will suppose Tk=k, where k - the term, expressed in
months.
Time tools (issued loans, bonds, interbank credits, time deposits) from the mathe-
matical point of view are similar, that is why we will consider them in the context of
one and only construction, giving the general designation: Xk - to time tools in assets
and Yk – in liabilities. Then the previous model can be presented as:
dX k
uk (t ) k X k xk (t , k ) uk (t ) k X k uk (t k ) exp( k k ) (13)
dt
dSt St dt St dWt f (t )dt (14)
Mathematical Model of Banking Firm … 407
dQ dY dX dZ
k k k X k k Yk g (t ) f (t ) (15)
dt k dt k dt dt k k
dYk
vk (t ) k Yk yk (t , k ) vk (t ) k Yk vk (t k ) exp( k k ) (16)
dt
dZ Z
w(t ) (17)
dt Dz
where w(t) - inflow of on-demand deposits , vk(t) - inflow of time deposits and bor-
rowed funds; f(t) - purchase (+) or sale (-) trading securities (t/s); g(t) – operation
costs on carrying out of activities of bank; - securities portfolio return; - volatility
of securities portfolio; Wt- Wiener stochastic process; k - interest on the time deposits
and borrowed funds; k - interest on issued loans; Dz - duration (characteristic turn-
over time) on-demand deposits.
It is easily to obtain the equation of dynamics of equity by differentiation of bal-
ance equality and corresponding substitutions (13) - (17). As follows,
dC dS
k X k k Yk t f (t ) g (t ) (18)
dt k k dt
For simplicity it is supposed complete withdrawal of deposits after term in this ver-
sion of model. However it is easy to take into account possibility of prolongation of
the deposit or its transfer in category on-demand deposits. It is considered that divi-
dends are not paid.
Besides, credit risks (default risk, or a delay of payments) are not considered, that
also it is possible to take into account by entering of corresponding adjustments.
It is considered that interests on the attracted funds and the received credits are paid
according to accrual. However it is easy to set and other scheme in which interests are
accumulated on depositary accounts and are paid after term of deposit.
Let k X k X and - k Yk Y - structure of time loans and deposits.
Besides, for simplicity we will assume that there are no investments in trading se-
curities. Then dynamics of the capitals are described by the equation:
dC
X k k Y k k g (t ) , (19)
dt k k
It is giving evident representation about sensitivity of dynamics of capital to
changes of main parameters of assets and liabilities.
Main objective of shareholders and bank management is the increase in capital:
dC
max (20)
dt
subject to restrictions on financial resources and risks (credit and market, loss of li-
quidity, bankruptcy).
408 V. Selyutin and M. Rudenko
4 Conclusions
The approach to mathematical modelling of cash flow moving in asset and liability
accounts of the commercial bank based on the partial differential equations is novel
and has no analogues in the literature. At the same time, the given approach is quite
logic as reflects process of change of actives simultaneously in time and on "age".
Depending on particular theoretical or practical problems the given approach can be
realized in the various modifications, two of which are presented in the article.
As the preliminary testing has shown, the computer program created by use model
(2-3) allows provide various simulations, including stress-testing, and can be used in
the educational purposes to provide the best understanding of the dynamic processes
taking place in banking firm.
It is necessary the further development of the offered modelling approach such as
improvement of program tool and also, as required, model detailed elaboration to use
these models as part of decision support system for asset and liability management in
commercial bank. The modified model (13-18) has been proposed for these goals.
References
1. Chi, G., Dong, H., Sun, X.: Decision Making Model of Bank’s Assets Portfolio Based on
Multi-period Dynamic Optimization. Systems Engineering – Theory & Practice, 27(2),
1–16 (2007)
2. Kruger, M.: A Goal Programming Approach to Strategic Bank Balance Sheet Manage-
ment. Banking, Financial Services, and Insurance. In: Proc. SAS Global Forum, Paper
024–2011 (2011)
3. Kosmidou, K., Zopounidis, C.: Asset Liability Management Techniques. Handbook of Fi-
nancial Engineering, pp. 281–300, Springer Science+Business Media, LLC (2008)
4. Mukuddem-Petersen, J., Petersen, M.A.: Bank Management via Stochastic Optimal Con-
trol. Automatica 42, 1395–1406 (2006)
5. Mulvey, J.M., Shetty, B.: Financial Planning via Multi-stage Stochastic Optimization.
Computers & Operations Research 31, 1–20 (2004)
6. Solyankin, A.A.: Computerization of the Financial Analysis and Forecasting in Bank. Fin-
StatInform, Мoscow (1998) (in Russian)
7. Robinson, R.S.: BANKMOD: an Interactive Simulation Aid for Bank Financial Planning.
J. Bank Res. 4(3), 212–224 (1973)
8. Moynihan, G.P., Purushothaman, P., McLeod, R.W., Nichols, W.G.: DSSALM: a Deci-
sion Support System for Asset and Liability Management. Decision Support Syst. 33(1),
23–38 (2002)
9. Alekseev, I.V., Selyutin, V.V.: Interactive Computer Model of Bank`s Asset and Liability
Dynamics. Terra Economicus 9(4), Part 2, 42–47 (2011) (in Russian)
2.2 1st International Workshop on Methods
and Resources of Distance Learning
(MRDL 2013)
Foreword
1st International Workshop on Methods and Resources for Distance Learning
(MRDL) has taken place on 19-22 of June 2012, in Kherson, Ukraine in conjunction
with 9-th International Conference on ICT in Education, Research, and Industrial
Applications: Integration, Harmonization, and Knowledge Transfer (ICTERI 2013).
Distance learning is an important application field that intensively uses information
and communication technologies in education. MRDL workshop will bring together
reports dealing with the problems of resource maintenance, pedagogic and didactic
methods of use of distance learning technologies.
The scope of the MRDL workshop includes the following topics:
Virtual laboratories: Mathematical and informational models and educational
tools for virtual labs, covering in particular the distance learning courses in phys-
ics, chemistry, biology and other disciplines.
Design and development of electronic learning tools and resources: Design,
development and use of electronic learning resources. Compatibility and integra-
tion of electronic resources in distance learning systems.
Computer-aided learning systems: Design, development and use of computer-
aided learning systems.
Pedagogical innovations in distance learning: Experience in developing and
implementation of new pedagogical methods of distance learning using ICT. Open
distance courses and tutors training in author's course content.
Monitoring of learning quality: Methods and tools for the estimation of quality of
knowledge in distance learning systems, testing, rating systems, feedback.
Quality of electronic learning resources: Standards for electronic resources for
distance learning. Modeling of and experience in using quality management sys-
tems for electronic learning resources.
Information of teaching and educational institutions: Modeling and experience
in the use of management systems in information processing and other manage-
ment processes at educational institutions. Experience in the use of distance learn-
ing technologies in a traditional educational process.
A blind peer-review process by at least two reviewers with expertise in the area has
been carried out. As a result, 22 submissions have been accepted as reports, 3 of them
are presented in this edition. We would like to thank the authors for their submissions
and our Program Committee members for their reviews.
June 2013 Vladimir Kukharenko
Yulia Zaporozhchenko
Hennadiy Kravtsov
What Should be E-Learning Course for Smart Education
Natalia V. Morze1 and Olena G. Glazunova2
1
Borys Grinchenko Kyiv State University
n.morze@kmpu.edu.ua
2
National University of Life and Environmental Sciences of Ukraine
e_glazunova@yahoo.com
Abstract. The article deals with problems of creation and use of e-learning
course for smart education. Structural features, the ratio of form and content of
the smart course elements and its properties: individual learning paths, content
personification, the use of training elements with links to public information re-
sources, interactive training elements, multimedia, communication and coopera-
tion elements are substantiated.
Keywords. Smart education, e-learning course, informal learning, individual
learning path, services of social networks, Content Learning Management Sys-
tems
Key terms. KnowledgeEvolution, KnowledgeManagementMethodology, Di-
dactics, KnowledgeManagementProcess, ICTInfrastructure.
1 Introduction
Modern information society is gradually transformed into Smart Society, as noted by
sociologists, philosophers, specialists in IT sector, educational specialists, etc. This
concept implies the new quality of society, in which a set of technological means,
services and Internet used by trained people, leads to qualitative changes in the inter-
action of subjects that allow receive new effects – social, economic and other benefits
for a better life [1].
During Smart Society formation the paradigm of education and educational tech-
nology is naturally changing. The tasks of training of the new format specialist, suc-
cessful and competent to work in the Smart Society rely on the new universities –
Smart Universities where the integration of technological innovations and the Internet
can provide a new quality of the educational and scientific processes, the results of
training, scientific, innovation, educational, social and other activities.
The conceptual basis of the Smart University is a large number of different scien-
tific sources, and information and educational materials, multimedia resources (audio,
412 N. V. Morze and O. G. Glazunova
graphics, video) that can be easily and quickly designed, assembled to a certain set,
adjusted individually for each student, his/her need and peculiarity of educational
activity and the level of educational achievements.
2 Problem
It is obvious that in conditions of development of Smart Society the educational para-
digm will also change. Smart Universities will perform new functions. Accordingly,
the requirements for e-learning courses that ensure students’ needs in educational
resources will change. Our mission is to substantiate theoretically the properties of
such e-learning course, its structure and components, and to test the effectiveness of
its use experimentally as well.
3 The Presentation of the Main Research and Explanation of
Scientific Results
3.1 Characteristics of the Smart University
5 key characteristics of the Smart University can be distinguished: social orientation,
mobility, accessibility, technological effectiveness and openness [2].
Social orientation consist in the personalization of education, building of the indi-
vidual education cards (Smart-card), organization of the efficient communication and
collaboration in education, cooperation, application of design and game techniques,
communication via social networks services, etc.
The second, equally important, feature of the Smart University is mobility. Mobil-
ity should be understood not only in the narrow interpretation - as an access to the
educational content through mobile devices and their use for scientific researches,
payment transactions, implementation of feedback with the teacher or the representa-
tives from the dean office or departments, etc [3]. Mobility is important as an access
of each student and teacher to the educational services from any place and at any
time.
Accessibility as feature of the Smart University is characterized by a single point of
entry to e-learning and scientific databases, media library, information kiosks, online
resources and access control systems to them, etc.
Technological effectiveness provides a viability of the Smart University IT infra-
structure by the means of cloud-technologies, innovative technologies of virtualiza-
tion, open interfaces, based on the principles of simplicity, modularity, scalability, etc.
Openness in the system of the Smart University foresees availability of the open
repositories of educational materials for forming e-learning courses and providing
training for students, open access to scientific articles and conducted researches and
their results [4].
What Should Be E-Learning Course for Smart Education … 413
3.2 Infrastructures of the Smart University
Modern university should have the appropriate infrastructure to support the require-
ments of the Smart Education. In particular, the activity of e-learning center, multi-
media center, scientific laboratories with the relevant open virtual environments and
open resources, library, including electronic one with the open access to the resources,
multimedia classrooms and computer labs should be based on the use of advanced
campus network with Internet access, including one on the basis of wireless technolo-
gies, cloud infrastructure, technologies of mobile access to e-learning resources, sys-
tem of distributed access to the resources. The effective functioning of such a sophis-
ticated infrastructure is impossible without a united center of data processing, from
where materials are distributed to the structural units: institutions, faculties and de-
partments, regional branches, dormitories, academic buildings, student centers, etc.
(Fig.1).
Effective activity of the Smart University will enable to realize not only the tasks
of formal, but nonformal and informal training as well. According to studies [5,6]
nonformal and informal learning takes 70% in the total structure of the educational
process and only 30% of learning is formal, in other words structured by years and
semesters of training, learning plans and programs that is usually provided by the
institution.
3.3 Features of the Smart Education
Smart Education sets a number of tasks for teacher on which performance depends the
effectiveness of teaching and students motivation to nonformal and informal learning,
which is based on the students skill to study independently. To interest the modern
student, who has access to a large number of high quality modern electronic materials
that can be easily found in the Internet, by conventional text linear (non-multimedia)
materials, only presented in electronic format, nowadays is almost impossible, espe-
cially in formal training. We should create such resources that will integrate multime-
dia, text, feedback tools on the basis of specific teacher’s individual recommendations
and external electronic resources that will meet individual needs and characteristics of
the modern student - regular user of the Internet resources and social networks. There-
fore the integral components of information and educational environment of the mod-
ern university should be: institutional repository of knowledge with full-text elec-
tronic educational and scientific resources; educational portal, which provides elec-
tronic support of all student’s learning activities for each discipline in the form of e-
learning courses with individual tasks and distinct and clear evaluation criteria that
are implemented with tools and methods of forming assessment; video portal with
multimedia resources for teaching and research activities; wiki portal as an environ-
ment to provide teamwork and collaboration; online services based on the use of Web
2.0 and Web 3.0 services and technologies, etc.
One of the main tendencies in the development of Smart Education is the openness
of learning systems – placing the educational content openly available to students
around the world, the development of systems with open code, development of
414 N. V. Morze and O. G. Glazunova
knowledge-sharing under the scheme "student-student", "teacher-teacher", "students-
teacher" and "students-teachers" [7,8]. An important step in the development of the
idea of massive open electronic courses was the adoption of the UNESCO declaration
on global policy on the issue of open e-courses, which sets the task of developing
standards for electronic courses, providing synergy in access to them, conducting
educational seminars on the development of courses and their use, collaboration be-
tween scientists and teachers, education quality assurance [9].
Fig. 1. SMART University infrastructure
3.4 Properties of E-Learning Course for Smart-Education
Some scholars define electronic course as didactic computer environment that con-
tains classified material from the relevant scientific and practical field of knowledge
that is combined by a single software shell, in which the following functional compo-
nents are selected: information and navigation (meaningful connections, annotation
and course structure, information, system of references, the searching system), infor-
mative (interrelated informative elements of the course – theory, practice, guidelines,
additional materials, information resources, including electronic and open), diagnostic
(formative assessment tools in the form of clear evaluation criteria for all types of
students activity, including self-assessment and mutual assessment, evaluation not
only of academic achievements of students, but also evaluation of formation of skills
of the 21-st century – to solve problems, work in team, communicate effectively and
collaborate, etc., the testing system of current, intermediate and final control) [10].
Electronic course for Smart education should provide flexible learning of the students
in an interactive learning environment, which allows him to adapt quickly to the envi-
What Should Be E-Learning Course for Smart Education … 415
ronment, to study in any place, at any time on the basis of free access to content all
over the world. In our opinion, the electronic course for Smart Education can be rep-
resented as a certain scenario or trajectory of educational events how to work with
electronic resources in the form of knowledge-map that leads to the achievement of
learning effect and has the following properties:
Flexibility – enabling rapid resources editing and making adjustments in educa-
tional trajectory
Availability of individual learning scenario, in other words, the possibility to draw
up an individual educational program for each student from the set of training ele-
ments
Integration of training elements with other open information resources
Focusing on the learning needs of the student, the personification of content
Interactivity of learning elements of the course, the maximum use of multimedia
technologies (videocasts, animation, video tutorials, screencasts, etc.)
Feedback between the teacher and the student in the course
Availability of training elements to ensure effective communication and coopera-
tion of students between themselves and with the teacher, in particular based on the
design technology [11]
Availability of game educational elements
Providing communication through modern services of social networks [12]
Creation of e-learning courses usually is carried out with the help of Content
Learning Management Systems. To create an effective e-learning course for Smart
education not only available electronic resources of information and educational envi-
ronment of the University should be used, but also open external information re-
sources and Web services that will serve as sources of educational and informational
materials for electronic course and as means of communication and cooperation
(Fig.2).
Information and educational environment of the university should be focused on
solving the problem of joint creation and use of academic knowledge for the needs of
students and teaching staff of the university. On the one hand, the teacher by himself
adds academic resources to the information and educational environment, such as
video clips and video tutorials posted on educational video portal and on the other
hand, he has the possibility to use available public resources for creating e-learning
course. So, to create an electronic course it is sufficient for the teacher to actualize
material that is available from other sources, submit it in accordance with the above
mentioned properties and criteria of evaluation of its quality, add the necessary train-
ing elements of the course according to the adopted structure and develop an individ-
ual learning scenario for each student, consider the individual evaluation criteria of
educational achievements of students and developed skills of the 21st century.
3.5 Structure of E-Learning Course for Smart-Education
Analysis of papers devoted to the creation and use of e-learning courses [13, 14] led
to the conclusion that in the issue of the course structure they should be focused on
the modular principle of its construction. When structuring the content of educational
416 N. V. Morze and O. G. Glazunova
subject by the principle of training modules each module should consist of intercon-
nected theoretical, empirical and practical components of the content, each of which
would carry out an independent function. Thus the educational discipline module is an
information and didactic unit, in which the approach to structuring the whole into
parts is unified. It has a complex structure that includes goal of its integral develop-
ment, objectives, content and results with the corresponding system of formative as-
sessment.
Fig. 2. Sources of electronic course formation for Smart education
Furthermore, the structure of the e-course for Smart education should provide
availability of:
Tools to build individual learning trajectory (prior surveys, questionnaires, tests,
formative assessment tools, including check-lists and tables of evaluation criteria,
etc.)
Multimedia presentations of summarizing character, video resources, interactive
electronic manuals, external Web resources with multimedia theoretical material
Links to external public resources including articles, conference proceedings, re-
search materials, etc.
Discussions on the forums, feedback with teacher, webinars and other Web ser-
vices
Intermediate control elements during the lessons and formative evaluation instru-
ments, final control in the form of control tasks and tests, element of reflection
Each element of the training course must meet certain standards and be evaluated
using criteria that are accepted at the level of educational institution [15].
Approximate structure of Smart course is shown in the Figure 3.
Example of the course topic, created in the CLMS Moodle environment, presented
on training and information portal NUBiP Ukraine
(http://it.nubip.edu.ua/course/view.php?id=21).
What Should Be E-Learning Course for Smart Education … 417
Further we will focus in more detail on the features of the e-learning course struc-
ture for Smart education.
Fig. 3. Structure of the electronic training course for Smart education
3.6 Formation of Individual Learning Trajectory
For the modern student, who has formed basic IT competences, there is a need not
only in the access to the resources, but mostly in the navigation knowledge-map,
"guidebook" to knowledge, that can be found in information space, as it is important
to help student to find quality resources. And this is a complex task for untrained
student. Smart education using Smart courses of the new model is the most comfort-
able and modern teaching model for such cyber-students. To build individual training
418 N. V. Morze and O. G. Glazunova
trajectory of the student in the electronic course you can use several approaches. One
of them lies in the prior survey and testing of the students in terms of competence in
the course educational material and the preparation of educational trajectory on the
results of such survey and their identified learning needs (Fig. 4). Thus the survey is
based on extensive use of formative assessment tools that provides self-assessment
and mutual assessment.
Fig. 4. Stages of the individual trajectory construction
During the experimental study of the introduction of e-learning course of the new
sample for the students of the "Computer Sciences" specialty a survey was conducted
for assessing their competencies on the subject under the scale: "have a good knowl-
edge", "be partially familiar", "heard something", "not familiar". Then each student
was offered a test for competence in the training material, which he/she "has a good
knowledge of" and "is partially familiar" with. According to the survey and testing
results individual learning trajectory was built for each student or group of students.
In other words, sequence of learning elements of the course was chosen, which stu-
dent should study. Moodle platform that we used to create the course allows to make
each training element available to a particular group of students. Therefore each stu-
dent or group of students receives an individual set of training elements of the course.
Training course is adapted for personal characteristics of each student that allow to
implement personally oriented approach and to develop an individual training pro-
gram. At the same time the course itself does not changed, but the methods of presen-
tation, set of tasks for performance and the tools, methods of evaluation and control
are changed.
What Should Be E-Learning Course for Smart Education … 419
3.7 Presentation Educational Material in the Theoretical Resources of Smart
Course
Peculiarity of the new model electronic course is also the diversity of the theoretical
learning resources presentation forms. Besides the theoretical material must be deliv-
ered by 60-70% in the multimedia interactive form, we also note the necessity to
choose the method of material delivery, depending on the level of its teaching. The
theoretical material in electronic course can be delivered on the following four levels:
phenomenological, analytical and synthetic, mathematical, axiomatic [16]. Each level
has its peculiarities in the delivery of educational material (Fig. 5).
Fig. 5. Ratio of levels and methods of educational material presentation in the theoretical re-
sources
Phenomenological level is characterized by the descriptive way of presenting edu-
cational material. Therefore, these materials should be delivered in the form of mul-
timedia presentations, interactive electronic manuals with graphics, multimedia and
video elements. Analytic and synthetic level is characterized by the necessity of pre-
senting of the theory of individual phenomena in naturally logical language that cre-
420 N. V. Morze and O. G. Glazunova
ates background for phenomena and processes forecast on a qualitative level. For this
level animation resources with elements of cognitive graphics should be prepared that
will be able to demonstrate the nature of the phenomenon and its dynamic changes.
Video tutorials with explanation and demonstration of the logic of the processes as
well as sound screencasts will be also effective. The mathematical level is character-
ized by the use of mathematical tools for modeling, theorem proving, examples of
solving problems, etc. Therefore, conventional textbooks are not enough to deliver
such material. It is necessary to create resources in the form of video lectures, video
lessons, and text resources should be reduced to the minimum amount – in the form of
handbooks with basic rules, formulas, theorems, etc. Educational material of the
axiomatic level can be presented in the form of video tutorials, e-manual and multi-
media presentations. Also it is necessary to actively use the links to external resources
that cover material from the considered topic. Such resources will add credibility to
the course and allow students to familiarize with additional sources of educational
materials.
3.8 Presentation Learning Tasks in the Smart Course
Another feature of the e-course for Smart education is the existence of elements for
communication and cooperation between the students in the performance of tasks of
mastering theoretical material, practical tasks, research projects, etc. Web 2.0 ser-
vices, online services, social networks provide tools for organizing discussions, col-
laboration, counseling. These elements are embedded in the course directly through
the platform that is used, or by reference to it. While performing tasks students should
use modern information and communication technologies effectively. Usually for
mastering the theoretical teaching material students (Fig.6) are offered tasks on
writing essays, composition of the related bibliography, writing summaries, para-
phrasing theoretical information of a small amount in the form of "question-answer",
compiling a glossary of terms under the certain topic, performing descriptive works,
making instructions for the implementation of various operations, plotting grid plans
and schedules.
In order that such tasks become interesting for students it is necessary to use Inter-
net resources, and present the result of the performance in the electronic form using
modern information technologies.
Tasks on mastering practical skills include solving problems, performance of ex-
ercises, graphical works, practical works, calculated works, designing, modeling,
compilation of practical situations from their own experience and on the basis of prac-
tical training, performing analysis of enterprise activity. Students should be offered to
solve such tasks using virtual laboratory workshops and specialized software.
Tasks on forming research activity include implementation of individual research
tasks, writing term papers, graduation works, participation in the educational projects.
Such types of tasks involve creative activity of the students, which should be carried
out by means of modern computer technologies, teamwork and communication. And
one of their peculiarities is the use of formative assessment tools for their assessment
that clearly guide the student to achieve learning goals in all types of educational
What Should Be E-Learning Course for Smart Education … 421
activity that are presented specifically, clearly and should be achievable for each of
them.
Fig. 6. Types of tasks to master practical skills
Thus, the main features of the students practical work organization using e-course
in the Smart education is the availability of tools for collaboration, communication,
combination of different information technologies in the performance of tasks. But
we should not forget that tasks should have practical significance, and contain de-
tailed information on their implementation, evaluation criteria and support resources.
We offer such pattern for the formulation of the task in the e-course (Fig.7)
3.9 Results of Experimental Research
In the course of research conducting we proposed a new model of electronic learning
course for students of shortened training period. As a result, after questioning and
testing six groups of students were identified who studied in different educational
trajectories, successfully completed training course and demonstrated 13% better
academic progress compared to the group of students who studied in one training
trajectory. At the same time, the participants of the experimental groups performed
larger volume of tasks on the depth study and worked extra theoretical study material
according to the references to the external information resources. In addition, teach-
ers-participants of the experiment, indicated that the presence of distributed environ-
ment of the opened resources in the university allows to create e-learning courses
applying much lesser efforts and requires lesser time compared to the case when the
422 N. V. Morze and O. G. Glazunova
course is created from the beginning. Teachers are able to use ready resources for
creating elements of the course – presentations, video recordings, electronic versions
of manuals and guidelines, a database of scientific publications, etc.
Fig. 7. Pattern for the formulation of the task in the e-course
4 Conclusions
Thus we note that the electronic course that has the properties required in the view of
Smart education is an effective tool for nonformal and informal learning, in which
most motivated students are interested now for obtaining high-quality knowledge, not
only a diploma of higher education. For efficient organization of learning activity in
the conditions of Smart education modern university should have distributed informa-
tion and educational environment that will enable to concentrate open electronic
learning resources and to move knowledge into a distributed network, actively use the
Web 2.0 services, mobile technologies, management system for learning content for
delivering knowledge to the students and the interactive exchange of information data
and training materials with them. In the future the development of such approach is
possible due to the joint development and use of the open educational content reposi-
tory by the universities based on the technologies of the Smart education.
What Should Be E-Learning Course for Smart Education … 423
References
1. Tikhomirov, N.V.: Global Strategy for the Development of Smart-Society. MESI is on a
Smart-University, http://smartmesi.blogspot.com/2012/03/smart-smart.html (In Russian).
2. Measuring the Information Society 2012, Committed to connecting the world,
http://www.itu.int/dms_pub/itu-d/opb/ind/D-IND-ICTOI-2012-SUM-PDF-R.pdf (In Rus-
sian).
3. Traxler, J.: The Learner Experience of Mobiles, Mobility and Connectedness. Evaluation
of Learners’ Experiences of e-Learning Special Interest Group.
http://www.helenwhitehead.com/elesig/ELESIG%20Mobilities%20ReviewPDF.pdf
(2010)
4. McAuley, A., Stewart, B., Siemens, G., Cormier, D.: The MOOC Model for Digital Prac-
tice. http://www.elearnspace.org/Articles/MOOC_Final.pdf
5. Kuharenko V. M.: Formal, Informal, Informalne and Social Studies. In: Modern Educa-
tional Technology in Education, pp. 114–124 (2012) (In Russian)
6. Mapping Informal and Formal Learning Strategies to Real Work.
http://performancexdesign.wordpress.com/2011/05/04/mapping-informal-and-formal-
learning-strategies-to-real-work
7. Helmer, J.: A Pair of Key Trends for this Year Learning: MOOCs and OA,
http://www.smart-edu.com/moocs-and-oa.html
8. Open Educational Resources, http://www.unesco.org/new/en/communication-and-
information/access-to-knowledge/open-educational-resources/
9. Pawlowski, J.M., Hoel, T.: Towards a Global Policy for Open Educational Resources: The
Paris OER Declaration and its Implications, White Paper, Version 0.2, Jyväskylä, Finland,
(2012)
10. Bezdolny A.V.: Model of E-Learning Course as a Means of Organizing the Self-Training,
http://cyberleninka.ru/article/n/model-elektronnogo-uchebnogo-kursa-kak-sredstva-
organizatsii-samostoyatelnoy-podgotovki (In Russian)
11. Gnedkova O., Lyakutin V.: Methodological Recommendations of Internet-Services Usage
in Distance Learning System “Kherson Virtual University”. Information Technologies in
Education, 10, 183–187 (2011) (In Russian)
12. Ravenscroft, A.: Dialogue and Connectivism: A New Approach to Understanding and
Promoting Dialogue-Rich Networked Learning. International Review of Research in Open
and Distance Learning, 12(3), http://www.irrodl.org/index.php/irrodl/article/view/934
13. Osin A.V.: E-Learning Resources in a New Generation of Questions and Answers,
http://www.ed.gov.ru/news/konkurs/5692#g10 (In Russian)
14. Mosher, B.: Five Myths About Informal Learning, http://www.smart-edu.com/stati-
korporativnoe-obuchenie/pyat-mifov-o-neformalnom-obuchenii.html
15. Morze N. V., Glazunova E. G.: Quality Criteria for E-Learning Courses. Information
Technologies in Education, 4, 63–76 ( 2009)
16. Deryabina G. I., Losev V. Yu.: Creating E-Learning Courses: Studies. Samara. Univers –
groups (2006)
TIO – a Software Toolset for Mobile Learning
in MINT Disciplines
Daniel Sitzmann1, Dietmar P.F. Möller1, Karsten Becker2 and Harald Richter3
1
University of Hamburg, MIN Faculty, FB Informatik, AB TIS, Building F,
Vogt-Kölln-Str. 30, 22527 Hamburg, Germany
{sitzmann, dmoeller}@informatik.uni-hamburg.de
2
Hamburg-Harburg University of Technology, Institute of Computer Technology,
Schwarzenbergstr. 95E, 21071 Hamburg, Germany
k.becker@tuhh.de
3
Clausthal University of Technology, Arnold-Sommerfeld-Str. 1, 38678 Clausthal, Germany
richter@tu-clausthal.de
Abstract. A web-based tool-set called TIO was created for mobile learning
with emphasis on study courses in mathematics, informatics, natural sciences
and technology (MINT). Mobile learning is a variant of E-learning which is
based on mobile user devices with Internet access. It was tested for numerous
software and hardware configurations users may have and proved to be techni-
cally working. TIO consists of a modified version of the open-source E-learning
system ILIAS and a tool set. The experience with TIO was that mobile learning
is useful for MINT subjects provided that numerous end user devices are sup-
ported and several text systems as well. We found that mobile learning is espe-
cially useful for professionals because they can learn in their free times in a
flexible way. Finally, we found that an emotional component should be existent
to make mobile learning more lasting.
Keywords. Mobile learning tool, MINT, single-source publication, social me-
dia
Key terms. KnowledgeEvolution, KnowledgeManagementMethodology, Di-
dactics, KnowledgeManagementProcess, ICTInfrastructure
1 Introduction
Mobile learning or M-learning is a new form of E-Learning. Mobile learning means
that pupils, students, or professionals are learning via mobile devices such as note-
books, E-books, handhelds, PDAs, smartphones, tablet PCs, iPads, iPods or gaming
consoles. The term MINT stands for mathematics, informatics, natural sciences and
TIO – a Software Toolset for Mobile Learning in MINT Disciplines 425
technology. The combination of both, i.e. the application of M-learning in MINT
subjects is not yet found in literature but addressed in this paper. Because there is no
software that supports M-learning for MINT subjects, it had to be developed. The
scientific questions of this project are: how should a tool-set look like that optimally
supports M-learning for MINT subjects, and what are the benefits and disadvantages
of teaching MINT subjects by means of M-learning in general. The latter question
will be answered in the future because TIO can be used for a subsequent pedagogical
evaluation of the effect and the adoption of M-learning in MINT subjects. It is the
technical basis and thus a prerequisite for assessing the benefits and disadvantages.
The first research question will be answered in the following because tool-set called
“Technical Informatics Online” (TIO) is presented which is a software platform for
computer-aided distance learning that is web-based and that has an emphasis on
MINT disciplines and on mobile learning. TIO is a front-end for authors of teaching
material, and at the same time it is a user interface for learners. It provides for both
groups spatial and temporal flexibility in creating content and in consuming it.
Basically, TIO is for the editing of teaching material, its distribution to various
mobile end-user devices, for managing study courses, and for learning these course
materials. TIO supports so-called single source publishing and serves as a social me-
dia for its users in order to make learning a deeper and thus longer-lasting experience.
Single source publishing allows to create, to maintain, to retrieve and to deliver the
very same content for many heterogeneous end user devices while it is stored only
one time in one file. In order to achieve this, a TIO-internal XML-based data structure
called “xml4tio” was defined that can be converted by TIO tools into various output
formats which optimally support the respective end user device. Because these de-
vices are very different in their capabilities, several presentation formats must be gen-
erated out of the same source file, depending on the user's preferences and device
type.
TIO is based on a modified version of the open-source software ILIAS and of a
TIO web application called TIOWA. It provides for numerous E-learning features,
including chat rooms, multiple-choice tests, and for the management of the teaching
content, the learning courses as a whole and their users.
The rest of the paper is organized as follows: chapter 2 describes the state-of-the-
art in M-learning, single source publishing and social. In chapter 3, An overview de-
scription of TIO is given that explains its software components and used technologies.
In chapter 4, the technical set-up of TIO is explained by means of block diagrams and
xml data structures in more detail. Also the made extensions to ILIAS are described
briefly here. In chapter 5, a report about field tests of the tool-set is given. The paper
ends with a conclusion, an outlook to future work and a reference list.
426 D. Sitzmann, D.P.F. Möller, K. Becker, and H. Richter
2 State-of-the-Art
2.1 M-Learning
M-learning is a modification of E-learning for the purpose of distance education and
blended learning based on mobile end-user devices. An overview on M-learning can
be found in [4], [5], [6] and [7]. Blended learning means that customers must show-up
in a classroom for a fraction of about 20% of their time and are not allowed to study
completely from remote.
In theory, M-learning has several advantages: First, mobile devices are already
widespread. Thus, no or little investment is needed for the user compared to a desktop
PC. This is important for many young clients and for customers from developing
countries. Second, mobile devices reflect the life style of the generation of ´digital
natives‘ [8]. Thus the potential acceptance and use of M-learning may be higher than
for classical learning styles. Third, mobile devices help professionals to consume
content on top of their working hours for the purpose of life-long learning because of
the access flexibility they get. They can use free time slots while travelling between
work and home for learning, or weekends and holidays in a very flexible way. Fourth,
Internet access allows quicker distribution of content and for significant lower media
costs compared to all other distribution ways that are based on paper. Fifth, if the
sensors of the end-user device such as the GPS position, the tilt- and acceleration-
meters are engaged then learning can adapt dynamically to the current location and
situation of the learner.
In practice, M-learning faces several problems: First, small screen and key sizes
hamper its applicability in all cases except of notebooks. As a consequence, a sophis-
ticated layout and formatting of teaching material becomes less important because
such formats may not be displayed. Additionally, limited user input must be tolerated
because of keyboard restrictions. Second, a rel. slow Internet access and a limited
battery life of the user device must be taken into account. Third, heterogeneous hard-
ware and operating systems with no or small hard disks are common. Mobile devices
have more options with respect to CPUs, RAM size, displays and operating systems
compared to desktop PCs. This makes it very difficult to present and use teaching
content equally on all devices.
Numerous E-learning platforms are already in usage. An overview can be found in
[9], for example. However, most of them are intended for use on desktop PCs and not
for mobile devices. Furthermore, their features for presenting technical content with
formulas for visualization of simulation results and for access to remote laboratories
is mostly limited. Finally, Internet access via mobile devices is not explicitly sup-
ported. TIO addresses these problems and supports text types that occur frequently in
MINT subjects.
2.2 Single Source Publishing
Single source publishing means that one internal storage format is used out of which
diverse customer outputs can be created, such as html with various cascaded style
TIO – a Software Toolset for Mobile Learning in MINT Disciplines 427
sheets and pdf. By single source publishing, a simpler updating process of content is
possible that allows to reuse the same teaching material for many devices and in vari-
ous contexts.
Traditionally, single source publishing is implemented by a 1:1 correlation be-
tween chapter and character formats in the source texts and, for example, the gener-
ated html code for the user. Additionally, source texts may be augmented with tags
and comments that give meta information about the text. From this semantic data,
converters can create automatically various output formats, provided that sufficient
meta data exists.
TIO uses a combination of both. First, it converts teaching materials that is format-
ted in a traditional way by chapters, sub chapters and emphasizes such as underline,
italic and bold into the common intermediate storage structure xml4tio. Subsequently,
it converts xml4tio into the needed user formats. Finally, it uses ILIAS and a self-
developed web application called TIOWA to disseminate the requested content in the
desired format via mobile Internet.
2.3 Social Media/Web 2.0
Social Media is a generalization of the term web 2.0. It denotes that users are not only
consumers of content via download but also producers via upload. It denotes further-
more that users are interacting with each other in the web. Examples of popular social
network services are Facebook, twitter, YouTube and Wikipedia. TIO strives to be a
social media for distant learning in the MINT disciplines in order to make learning a
deeper experience because the distinction between learning and leisure are more
flexible then, and because learning gets hereby an emotional component that exists
also in a classroom but not in a computer system.
3 Description of TIO
TIO combines E-learning via the Internet with mobile communication. It allows thus
for learning from anywhere and at any time. TIO is focused but not limited to mobile
user devices. It can be used also with desktop computers. Content creation can be
accomplished by authors via the commercial text system “Adobe Frame Maker” [1] if
technical manuscripts have to be written. A TIO tool converts these Frame Maker
texts in xml4tio. If the editing of mathematical texts is required then a self-developed
tool called LearnDSL (Learn Domain Specific Language) can be used for the conven-
ient writing of formulas in a style similar to LaTeX but more user-friendly. LearnDSL
converts mathematical texts into xml4tio. Finally, Open Office Writer [2] is supported
by TIO for all other use-cases. A macro-based converter transforms Open Office texts
again in xml4tio.
TIO is based on the wide-spread ILIAS E-learning platform [3] for which several
adaptions and extensions were developed. ILIAS implies PHP [10] as programming
language and SCORM [11] as a content format. TIOWA in turn allows to upload
teaching material and to convert it from xml4tio into SCORM and html. TIOWA uses
428 D. Sitzmann, D.P.F. Möller, K. Becker, and H. Richter
Ajax [12] which provides for the students to interact with TIO as if it were a local
application running on his hand-held device. Ajax employs several technologies such
as Javascript [13], XML [14], HTML5 [15] and CSS3 [16]. An important part of
TIOWA are several converters which are based on XSLT [17]. In principle, TIOWA
and ILIAS could even be used independently from each other for the creation, storing
and dissemination of teaching material. For example, TIOWA could also be con-
nected to other learn management systems such as Moodle [24].
Aims and Features. Our extensions to ILIAS provide for the following features:
First, authors can create teaching material in that text system that is preferable for
them and their content. After content creation, support for an automated conversion
into the intermediate xml4tio file format is provided by TIO. Xml4tio is an xml
schema definition that is used to store content on a TIO server [25].
Second, formatting instructions stored in xml4tio is reduced to a minimum because
small screens can not display sophisticated presentations that were created in Power
Point, for example, in a satisfactory manner.
Third, the use of the common storage format has the potential for easier access and
for better maintaining a large amount of teaching material, as it is needed for a Bache-
lor or Master courses, for example.
Fourth, TIO supports multiple end-user devices and operating systems by engaging
browsers as the only needed software at the end-user device.
Fifth, TIO supports the learning process not only for full time students, but also for
part time customers, together with training on the job for life-long learning of profes-
sionals. Therefore, it can be used by universities and enterprises.
Sixth, the teaching material is structured by TIO into ‘learn objects‘ which can be
compared to book chapters but may contain audio, video, text, graphics, formulas,
visualization of simulation and access to remote laboratories, as well as exercises and
multiple choice tests for each topic.
Seventh, TIO can administer a large number of students as customers and a large
number of courses as well because it uses ILIAS which has proven to be a reliable
system.
Eighth, most of the TIO portals are multilingual, currently in German and English,
which gives also a transnational aspect.
4 TIO Set-up
TIO‘s set-up is based on three (virtual) servers, one for authors, one for students, and
one for backup (Error! Reference source not found.).
TIO – a Software Toolset for Mobile Learning in MINT Disciplines 429
Fig. 1. General hardware setup of TIO
All servers use UBUNTU Linux. On the author‘s server, our TIOWA web application
is the main software component, together with several converters between storage
formats. TIOWA is the frame in which authors create, maintain and convert learn
objects. Additionally, a relational data base is provided that cares for authentification
and authorization of authors and system administrators, and that maintains different
versions of teaching objects. Teaching objects are stored as files in xml4tio. In Error!
Reference source not found., the software set-up of the author server is shown. On
the customers server, ILIAS is the software which cares for student administration
and examination, inter-student communication and content dissemination. In Error!
Reference source not found., we see the block diagram of the customer server.
Fig. 2. Set-up of the customer server
430 D. Sitzmann, D.P.F. Möller, K. Becker, and H. Richter
C us tomer S erver
T IO -IL IAS P latform T IO E x tens ions Multimedia
C ours es , web 2.0, forums , notes ,
S tudents , T es ts , c hats , ques tions , etc .
Mobile Devic es Audio &
T emplate S etting s for us ability, Video
adaptivity, interac tion Databas e
xml4tio2s c orm
Us er & C ours e Dis play and s earc h T IO
Databas e c ontent
L inked to
F rom
Multimedia
Data
T eac hing Material in S C O R M: HT ML 5/C S S
for S tationary or Mobile Devic es and P DF
T o C us tomers
Fig. 3. Set-up of the customer server
The information which student is enrolled in which course is stored in a MySQL
data base. Audio and video recordings of teaching objects and whole lectures are
stored as standard files with references to xml4tio and SCORM data structures.
SCORM and xml4tio are both based on xml. SCORM is a collection of specifications
and programming interfaces. It stores user preferences, records learning goals, logs
learn progress and describes which resources a learn object has. A resource is a set of
texts, pictures, audios, videos and URLs that is organized as a tree. Each tree resem-
bles the chapters and subchapters of a traditional lecture of 45-90 minutes and can be
augmented by meta data that describe the learn object. This meta data allows to search
for content. Furthermore, possible sequences can be specified in which students can
consume learn objects. Finally, the path through a sequence of teaching objects can be
declared as a function of the answers the user gives in multiple-choice tests. The
backup server provides for a safe operation by copying automatically data from the
author and customer servers. In the following, all described components are explained
in more detail.
4.1 The TIOWA Web Application
For content creation, management and format conversion, TIOWA was programmed
in PHP 5.4 and Javascript as a web application for the authors. It uses MySQL 5.6
[19] as data base for authors and administration, jQuery and Ajax for easier commu-
nication with the user via dynamic web pages, and an Apache web server for page
generation in HTML5 and CSS. TIOWA allows to upload content from an author‘s
computer to the TIO server, to convert it into xml4tio, to update and backup it, to
TIO – a Software Toolset for Mobile Learning in MINT Disciplines 431
move it to the customer server and, in the final step, to reformat the content for dis-
play, depending on the respective output device. For this purpose, TIOWA accesses
xml4tio data, author/admin data and SCORM data. From the viewpoint of xml4tio,
TIOWA is an application for single source publishing.
4.2 XML4TIO for Single Source Publishing
Xml4tio is a xml schema definition [24] and the core of TIO. For the conversion of
teaching material into xml4tio, two files and one extra folder must be created: First, a
´container.xml´ file is established that stores a description of the teaching material as
meta data. In case of B/M modules, the meta data are based on the Bologna module
description [20]. Otherwise, the author must provide an arbitrary text as abstract. Ad-
ditionally, container.xml binds together all teaching objects into lectures and all lec-
tures about the same topic into one Bologna module. Second, a ´content.xml´ file
must be created that stores the content of the Bologna module.
The first layer of the syntax tree of xml4tio comprises the xml tags module, title,
author, section and presentation unit. These tags can subsequently be specialized by
additional tags and/or attributes in more layers. Especially remarkable is for example
the ´media´ tag that can have as attribute ´picture´, ´animation´, ´applet´, ´scene3d´,
´sound´, ´video´ and ´experiment´. This offer shows the capability of TIO in media
presentations for MINT subjects.
4.3 Converters for Storing and Disseminating Teaching Material
There are converters from OO Writer, Adobe Frame Maker and LearnDSL into
xml4tio. A reverse conversion for a so-called round trip is normally not possible be-
cause nearly all character and chapter formatting instructions are deleted during the
conversion process. However, if the author limits himself in his text system on the
few formatting data that xml4tio has then he can also perform a roundtrip.
Additionally, there exists a converter for transforming xml4tio into SCORM, html5
and pdf. These formats contain less formatting information than the authors‘ original
documents.
The main difficulty the converters are facing with stems from the fact that humans
normally do not treat their lecturing materials as a formal text. This means, that beside
the used character and chapter formats, several hand-made changes exist in real-world
lectures. Such documents can not be converted. To detect these flaws, a xml4tio vali-
dator is provided that checks whether the lecture is structured according to the
xml4tio schema definition.
Conversion of Teaching Material into XML4TIO. For the LearnDSL converter,
own software in Java was written. For the OO Writer, a PHP software was created
that transform valid OO texts into xml4tio. Prerequisite for that is that a prescribed
LearnDSL and OO character and chapter format catalogue is used without any man-
ual additions or modifications. Furthermore, specific rules have to be obeyed with
respect to the structure of the teaching material.
For the Frame Maker (FM) converter, basically the same restrictions hold as for the
432 D. Sitzmann, D.P.F. Möller, K. Becker, and H. Richter
Writer converter. However, no software has to be programmed as in the LearnDSL or
OO case. FM must only be set from unstructured to structured mode. Then it can di-
rectly deliver the desired xml schema. FM uses as input for this schema a so-called
Element Definition Document (EDD) file [21]. This means that the EDD let Frame
Maker know which Elements are allowed and which composition of elements are
legal. The EDD additionally prescribes how to format them elements. By means of a
proper EDD, a structured text appears to the user of FM nearly as an unstructured
text.
The EDD in turn can be created out of a so-called document type definition (DTD)
[22] in a two-step manner: first the DTD file must be manually created on basis of the
xml4tio schema definition. Then, the DTD can be imported into FM, and an EDD is
created automatically by FM.
However, the resulting EDD file must be post processed manually to define the ap-
pearance of chapter and character formats of the elements. Such format definitions are
made by templates. The manual post processing of the EDD file is also needed to
provide for so-called read/write rules which help to convert FM documents into cor-
rect xml4tio by transforming FM elements into TIO tags and attributes.
After that, the EDD file must undergo an automatic transformation by XSLT to
create proper URLs, names and paths for pictures and other multimedia content ac-
cording to the xml4tio schema definition. Therefore, an XSLT style sheet was created,
and additional EDD rules were defined that call the style sheet.
Finally, FM bundles the DTD together with the EDD that contains format tem-
plates and read/write rules into a so called structured application definition. As soon
as such a definition exists, the user can just save any structured FM document into
xml4tio by simply clicking the "save as XML" button, and the conversion is done.
Conversion of XML4TIO to SCORM Output. This converter is called
´xml4tio2scorm´. It is responsible for content presentation because the created
SCORM file contains html for a browser, together with CSS formats and links to
multimedia data such as pictures and graphics. However, the full potential SCORM
has is not needed here. Only a TIO-specific subset is used that is subsequently stored
in ZIP format in order to save (virtual) disk space. xml4tio2scorm creates a table of
content for the selected study course, its html/CSS and pdf representation and a index
for searching. This index is copied automatically as a file into the user space of ILIAS
so that students can access it.
Depending on the fact whether the target browser is located in a mobile or station-
ary device and with respect to the device‘s screen size, either the table of content is
presented simultaneously together with a video recording of the lecture and the text of
the teaching material. Or in case of small screens, only one of these three streams is
displayed, according to the user wishes. Additionally, it is taken into account that
mobile devices may have only limited bandwidth for Internet access. Because of that,
offline browsing of previously downloaded teaching material is supported, together
with pdf viewing.
TIO – a Software Toolset for Mobile Learning in MINT Disciplines 433
4.4 Extensions to ILIAS
The following extensions to ILIAS were made: 1.) students can highlight and com-
ment every line of the teaching material. These comments can be attributed to be ei-
ther private, visible to the public or open for discussion in a special forum. This
makes students to prosumers. 2.) students can ask the author of the teaching material
via the Internet by placing their question directly into the script at the proper line
which eases the communication between student and teacher/author significantly. 3.)
Searching in the teaching material is possible by means of an index. 4.) Fonts, font
sizes and colors can be configured individually to provide for better reception. 5.) A
table of content can be displayed that helps to get a better overview of the learning
material 6.) The SCORM output is adapted to mobile devices with individual screen
sizes.
The extensions make students into “prosumers” that contribute to their learn suc-
cess by own comments and hints in the teaching material that are visible to all others.
The improved questioning and chat room feature allows a „learning adventure“ and
makes TIO into an Internet-based social media.
5 Software Tests and Practical Experiences
For testing the software we received from Clausthal University of Technology three
virtual servers which were also administrated by them, including backup and restart in
case of system crash. On these servers, three web portals were created: 1.) A portal
under http://webadmin.ti-online.org for accessing the TIOWA web applications. This
portal is for authors and admins. 2.) A portal for general information about the TIO
project and for project protocols under http://ti-online.org. 3.) A portal for TIO stu-
dents and other customers to download learning content under http://ilias.ti-
online.org/. This portal is also for teachers to manage TIO students and courses. It is
based on a modified version of ILIAS as described. All portals are maintained by
Typo3 [23] as content management system, the last portal is in German and in English
language.
5.1 End-User Devices and Configurations
TIO was tested with the following end-user devices and configurations: 1.) desktop
PCs and notebooks with MS Windows XP/7/8 as operating systems, and with Firefox
(>V2), Internet Explorer (>V6), Opera (>V8), Chrome (>V16) and Safari (>V5) as
browsers. 2.) desktop PCs and notebooks with Ubuntu, Debian and Suse Linux to-
gether with Opera and Firefox. 3.) Apple computers with Mac OS X (>V10.6), Safari
and Firefox. 4.) Apple iPad with iOs (>V5) and diverse Android Tablets (>V4.0). 4.)
Apple iPhone (V4) and diverse Android Smartphones (>V2.0), and 5.) Blackberry
and Apple iPod. This was considered a comprehensive selection.
434 D. Sitzmann, D.P.F. Möller, K. Becker, and H. Richter
5.2 Practical Experiences
In the years 2010-2013, teaching materials were created for TIO in Frame Maker,
LearnDSL and Open Office Writer. This happened for the education in Technical
Informatics, Computer Organization and Computer Networks at the Universities of
Hamburg and Clausthal. Additionally, some video recordings of the lectures, simula-
tions and animations are added to the teaching materials. After every semester, feed-
back from the students was collected in written form and evaluated. According to that,
parts of the software and the didactic presentation of the teaching material were im-
proved.
6 Conclusion and Future Work
A web-based software platform called TIO for mobile learning with emphasis on
study courses in mathematics, informatics, natural sciences and technology was cre-
ated. It was tested for numerous software and hardware configurations users may have
and proved to be technically working. TIO consists of a modified version of the open-
source E-learning system ILIAS and the TIO tool set. The ILIAS extensions were
made to improve its usability for mobile devices and MINT subjects. While ILIAS is
for students, TIOWA is for the authors for content creation, uploading and converting.
It stores the teaching material in a xml schema definition out of which several output
streams can be generated that depend on the end-user device and its screen size (sin-
gle-source-publishing). Teaching material can contain texts with formulas, audios and
videos, animations and visualization of simulation results. Finally, the xml4tio data
structure also allows remote access to experimental labs located in a university for
practical training. We believe that some of our implemented features will be useful
for the generation of ´digital natives´ that must prosecute life-long learning.
The general experience is that mobile learning is useful for MINT subjects pro-
vided that diverse end user devices are supported because they differ a lot in their
capabilities. Furthermore, it proved to be important to support various text systems
such as Frame Maker, Open Office Writer and a variant of LaTeX because otherwise
not enough authors can be found to develop teaching material. Furthermore, we found
that mobile learning is especially useful for professionals because they can learn now
in a flexible way in empty times slots such on train trips between work and home.
Finally, an emotional component should be existent in mobile learning that makes it a
deeper and more lasting. Thus chat rooms and other features with which clients can
communicate with each other in the style of a social media were integrated. As a rec-
ommendation, we suggest to combine mobile learning with classroom presence of
approx. 10% of the study time to make the emotional component optimal.
In the future, TIO must be approbated, tested and checked in practice over a longer
period of time. Therefore, it will be used for online teaching a Master course in Tech-
nical Informatics and for training and certifying of professionals for the purpose of
life-long learning. These real-world applications will allow to conduct a pedagogical
evaluation of the benefits and disadvantages of M-learning for MINT subjects. From
TIO – a Software Toolset for Mobile Learning in MINT Disciplines 435
that experience, recommendations can be given to teachers and organizations about
M-learning in general. The TIO tool-set is the technical prerequisite for that.
References
1. Adobe Framemaker, http://www.adobe.com/products/framemaker.html
2. Open Office Writer, http://www.openoffice.org/
3. ILIAS Learning Management System, http://www.ilias.de/
4. Holzinger, A., Nischelwitzer, A., Meisenberger, M.: Lifelong-Learning Support by M-
Learning: Example Scenarios, eLearning Magazine, 11, ACM, New York (2005)
5. Guy, R.: Mobile Learning: Pilot Projects and Initiatives, Informing Science Press (2010)
6. Kitchenham, A.: Models for Interdisciplinary Mobile Learning, Delivering Information to
Students, IGI Global (2011)
7. Sampson, D. G., Isaias, P, Ifenthaler, D., Spector, J. M.: Ubiquitous and Mobile Learning
in the Digital Age, Springer New York, Heidelberg, Dordrecht, London. (2013)
8. http://en.wikipedia.org/wiki/Digital_native
9. http://www.techworld.com.au/article/223565/10_open_source_elearning_projects_watch/?
10. PHP - Hypertext Preprocessor, Web Scripting Language, http://php.net/
11. Sharable Content Object Reference Model (SCORM), http://scorm.com/scorm-explained/
12. Ajax (Asynchronous JavaScript and XML),
13. Javascript, http://www.w3.org/standards/webdesign/script.html
14. Introduction in Extensible Markup Language (XML), http://www.w3.org/XML/
15. World Wide Web Consortium. Standards, http://www.w3.org/standards
16. World Wide Web Consortium. Cascading Style Sheets, http://www.w3.org/Style/CSS/
17. World Wide Web Consortium. XSL Transformations, http://www.w3.org/TR/xslt
18. World Wide Web Consortium. XML Schema, http://www.w3.org/XML/Schema.html
19. MySQL 5.6 Reference Manual, http://dev.mysql.com/doc/refman/5.6/en/index.html
20. Bologna process, http://ec.europa.eu/education/policies/educ/bologna/bologna.pdf and
http://www.ehea.info/
21. Element Definition Document, http://help.adobe.com/en_US/FrameMaker/8.0/help.html?
content=Chap2-FrameMaker-Basics_098.html
22. Document Type Definition, http://help.adobe.com/en_US/FrameMaker/8.0/help.html?
content=Chap2-FrameMaker-Basics_098.html
23. Typo3 - The Enterprise Open Source CMS, http://typo3.org/
24. Moodle - https://moodle.org/
25. XML4TIO, http://www.ti-online.org/XML4TIO
Holistic Approach to Training of ICT Skilled
Educational Personnel
Mariya Shyshkina
Institute of Information Technologies and Learning Tools
of the National Academy of Pedagogical Sciences of Ukraine
marple@ukr.net
Abstract. The article intends to explore and estimate the possible pedagogical
advantages and potential of cloud computing technology with aim to increase
organizational level, availability and quality of ICT-based learning tools and re-
sources. Holistic model of a specialist is proposed and the problems of devel-
opment of a system of methodological and technological support for elaboration
of cloud-based learning environment of educational institution are considered.
Keywords. Learning environment, personnel training, cloud computing, holis-
tic approach
Key terms. KnowledgeEvolution, KnowledgeManagementMethodology, Di-
dactics, KnowledgeManagementProcess, ICTInfrastructure
1 Introduction
As it is now impossible to introduce advanced ICT while managing this process with-
out mastering the ICT and other related pedagogical technologies, the main aim is to
train ICT-skilled educational personnel. Cloud computing technology (CC) is to cre-
ate a high-tech learning environment of educational institution, enhancing multiple
access and joint use of educational resources at different levels and domains. On this
basis it is possible to combine corporate resources of the university and other on-line
resources, adapted to learning needs, within a unite framework.
Cloud computing is used for resources supply and to support collaboration in the
learning process in particular by means of mobile services. It requires the develop-
ment of new approaches and models for designing of a learning environment. Among
them there are those based on a holistic approach to learning [1], [5], [7], [9].
For this aim a set of instrumentation tools for cloud-based learning resources col-
lection, elaboration and design, holistic models of learning environment and specialist
models, and a system of methodological and technological support for the develop-
ment of cloud-based learning environment of educational institution should be cre-
ated.
Holistic Approach to Training of ICT Skilled Educational Personnel 437
The purpose of the article is to identify trends and conceptual models of educa-
tional personnel training within the cloud based learning environment.
2 Problem Statement
The problem of training of qualified educational management personnel as well as
teachers oriented on ICT based learning can nowadays hardly be taken independently
from the processes of the innovative development of educational space formed within
the school, region and educational system of a country or globally [1]. In this regard,
there is a need for fundamental research focusing on the possible ways of developing
an educational environment of educational institutions. It should take into account, the
trends of improving ICT facilities while searching for new engineering technological
decisions and new pedagogical and organizational models [1], [2]. The main focus is
on shifting from mass introduction of separate software products, to an integrated and
combined environment which supports distributed network services and cross-
platform solutions.
Emerging technologies of information and communication networks give a way for
implementing a holistic approach to education and training of personnel. A holistic
approach focuses on combining science and practice, training and production, funda-
mental and applied knowledge and technological competencies with social and hu-
manitarian [5], [9]. Above all it aims the development of public administration’s
management skills in the educational field basing on a unite approach to learning
design and management. This is a promising direction for the development of a
field’s human potential. The innovative processes therefore, of the organization and
development of learning environment, search for new approaches and models for
specialist education and training becomes a matter of interest [11].
There is a problem of availability and valuable ways of learning resources deliv-
ery, to achieve with their use the best pedagogical effect and to gain maximum learn-
ing potential of ICT. This issue, may be hence partially solved if delivered by means
of cloud computing technology [2], [8], [12]. The main advantage of this technology
is the improved access to qualitative resources (and sometimes the only possible ac-
cess to necessary recourses at all). The idea is simply to explore approaches for the
modeling and estimation of CC-based learning process settings and valuable tools for
its organization.
3 Education and Training of ICT-Skilled Management
and Public Administration Personnel
Public administrators are public servants working in public institutions, departments
and agencies [10]. Specifically, they are concerned with “planning, organizing, direct-
ing, coordinating, and controlling government operations” [6]. Specific sphere are
public servants for education management. For such personnel to be efficiently
trained, the need to develop novel approaches arises, as this sphere is mostly con-
cerned with multi-disciplinary knowledge and requires skills on the merge of training,
438 M. Shyshkina
learning and management. Due to the fact that most pedagogical innovations are also
based on ICT the need in the sphere of education management also arises. There is a
branch of pedagogical sciences dealing with theoretical and methodological problems
of ICT in education use, psychological and pedagogical substantiation of these proc-
esses, elaboration of ICT tools and resources for providing functioning and develop-
ment of educational systems. So there should be specialized personnel to insure the
processes of implementation, introduction and development of ICT-based learning
technologies within the sector of public administration.
There are significant needs in IT competent specialists in the sphere of public ad-
ministration. Without ICT competence or competence in ICT for learning, problems
with their adaptation at the workplace arise, as do problems with the necessity of ad-
ditional and often profound training almost immediately after hiring. In some cases, a
vague idea of future graduates about the real problems and conditions of work with
innovative ICT infrastructures and ICT-based tools leads to lack of commitment to
practical solutions of work situations thus to a low level of innovative inclusion.
Formation of the innovative institution’s ICT infrastructure could solve some of
the aforementioned problems [11]. Namely, it would bridge the gap between the
process of training and the level of demand for their product. An environment that
would bring together the learning resources of educational and industrial projects
would be created, and would cover different levels of training; including the training
of both students and pedagogical management personnel.
According therefore to the high rates of development of both the global ICT mar-
ket for the education sector, and the IT market of learning tools, the problem of train-
ing professional staff for the domestic public administration and the IT-oriented sector
of education management; personnel which are primarily prepared within higher and
post graduate schools (e.g. universities and advanced training schools) being continu-
ous, we conclude that modern approaches to the design of educational systems are a
key point.
It is unlikely that the current state of skilled personnel of management and public
administration of education could be regarded as fully satisfactory for the needs of
innovative development of ICT-based learning, for the required number of qualified
professionals with appropriate structure and quality of training. The system of training
and retraining of employees for public administration has not been properly formed.
These problems should be considered within the context of development of an in-
stitution’s and a region’s innovative environment as well as on national and interna-
tional level [3], [11]. These processes have to do with the modernization of a learning
environment in perspective of emerging ICT. Thus the developmental need of new
models and approaches to personnel training arises, which will account for the mod-
ernization of ICT infrastructure and integrate resources of different levels and use.
Introduction of innovations into the educational environment of a state or a region,
is highly concerned with the development of human resources of informatization on
education [1]. It requires new types of skills and competencies which graduates often
lack of. These skills include leadership, ability to approach a problem holistically, and
the ability to critically evaluate achievement and self-assessment [9], [11]. It is the
lack of qualified personnel and the absence of a strategic approach to ICT infrastruc-
Holistic Approach to Training of ICT Skilled Educational Personnel 439
ture design that are among the reasons for an institution’s of professional education
deficiency of a unite high-tech desisions.
Nowadays content-technological process regarding the creation and use of ICT
products, and in particular the electronic learning resources, requires fundamental
background knowledge in both ICT and pedagogy. The approaches however for train-
ing personnel today, do not sufficiently take into account the recent years’ innovative
changes in the ICT industry, nor the real needs regarding the extent of such training.
A mean for provision of users with relevant services of cloud computing technol-
ogy is considered to be outsourcing; i.e. a service in a specific system to implement its
core functions is required, offered and sold by another system external to this [3]. ICT
outsourcing plays an important role in enhancing the scientific and technical level of
ICT-systems of an educational institution as well as the efficiency of their operation
and their development. It is a market mechanism incorporating the latest advances in
the ICT sector and to satisfying user demand [2], [3].
The main problem in educational practice is the contradiction between on one hand
the objective need for a continuous improvement of the software and the hardware
power of training computer complexes, and on the other, the lack of personnel’s abil-
ity (in both qualitative and quantitative manners) to maintain, manage and develop
their ICT systems appropriately. The informatization hence of an educational institu-
tion in terms of cloud computing and ICT outsourcing, will offer realistic solutions
for both the deepening of informatization and improvement of ICT’s educational per-
formance and use of information resources [2], [3].
The basic principles of such introduction should be: a tight relationship of learning
with training and methodological support for tutors, focus on a specific educational
task; modularity of learning; continuity of learning, sharing experience and formation
and participation in professional association activities (including electronic) [2], [3].
In this process electronic distance learning systems should be actively used, based on
the principles of open education, with the maximum possible use being of CC tech-
nology and outsourcing.
4 What are Advantages of Cloud Computing Decision?
4.1 It is a Cost Effective Solution
Being cost-effective the user can get (buy) products and services proposed by the
virtual supermarket of ICT according to their needs (individual or group, collective,
corporate), they may pay only for what has been bought (e -transport, e-content, e-
services, virtual e-tools, a generic and subject software applications, network plat-
forms - full range of cloud services along with services for the design and implemen-
tation of ICT systems and their fragments ordered by the users, their warranty and
post warranty service, maintain, upgrade and improvement, etc.) and only for the
actual time of use of the purchased product [3 ]. This will allow users to avoid regular
updating and upgrading of powerful general system software and hardware tools of
their own ICT systems, avoiding a potential surplus of ICT products used from time
440 M. Shyshkina
to time; fragmentary, not fully, as well as spare parts, reduction of requirements for
information security of their own ICT systems, reduction of the number of their ICT
services and requirements for professional competence of their employees and as a
result, significantly reduce overall costs to support the operation and develop their
ICT systems, to increase their social and economic return, their efficiency [2], [3].
4.2 This is a Flexible Solution of ICT Infrastructure
It is designed for increased flexibility and effective access to learning resources so as
to build a unified and mobile infrastructure.
On the basis of CC infrastructure all main aspects of interaction of a learner may
be comprehended on the unite basis. Along to approach introduced in [1], among
them there will be interactions between a learner and other learners; a learner and a
teacher; a learner and a learning tool; a learner and educational institution; a learner
and the society. This will lead to an environment of learning organization on the unite
base, where collaboration between learners and a tutor, free and flexible resource
access, learning activity within social inclusion into the environment of an educa-
tional institution and the society will be enabled. The ICT support of learning is real-
ized by means of cloud services. It is designed for adaptation to the rapidly changing
external/internal environment, changing of task/competence requirements and devel-
opment of modern pedagogical approaches.
Due to the principles of open education [1], there is a need to create an innovative
learning environment that will form and develop necessary professional skills. Among
them are leader skills, collaborative skills, critical thinking, and the ability to view a
problem in a holistic manner. These skills refer mostly to the demand of the sphere of
public administration of education, as in alliance with them; a process of innovative
development may be involved. This may be achieved on the basis of a holistic ap-
proach to specialist training when the planning, design and resource management and
learning activity of an organization and its monitoring, may be represented on a unite
basis. It will be achieved through the unite development of different competencies:
professional, fundamental, personal and technological.
5 A Holistic Model of a Specialist
Holistic approach to education deals with the learning processes to be taken as unity
of all main aspects of a personality development, for example such as mental, emo-
tional and volitional. This is in tune with a meaning of the term “holistic” as com-
pleteness, being impossible with disregard of some of its components.
There are innumerous investigations devoted to the problems of holistic learning
development in different aspects such as learning and teaching interaction, collabora-
tion processes, engagement of both aspects of theory and practice to gain comprehen-
sive view of a subject [7], [9]. Now there are important trends of research develop-
ment in concern to modern ICT. For example, holistic view is to approach learning
environment structure. Thus, the model of a learning environment, developed in [1] is
Holistic Approach to Training of ICT Skilled Educational Personnel 441
to reveal main components and types of interactions within the different learning
process settings.
The notion of holistic learning occurs in relation to personnel training, concerning
to different components and interactions within educational organization. It may
touch upon certain types of activity, collaboration and resource management proc-
esses, engaging thus the entire organization at all levels and developing a performance
culture of personnel. There are different ways to approach peculiarities of specialist
formation, namely in the aspect of personal or professional features. That concerning
to modeling of professional competencies [5], especially in the sphere of educational
management. Another aspect is about holistic models to develop leader skills [4], [9],
which are more to traits of a personality.
The proposed approach is based on holistic model of a specialist in the sphere of
informatization of education presented in Fig.1. It concerns to Domain Competencies
which would occupy fundamental knowledge of educational management and modern
learning technologies and also ICT skills and ability to use e-learning tools. There are
also Personal Competencies, such as leader skills, critical thinking, and capability to
holistic view of a problem, responsibility and activity of an individual. As for profes-
sional skills there are planning, design, resources management, cooperation and col-
laboration skills, performance skills and ability for monitoring and self evaluation.
Levels of Domain Competencies
Education
Fundamental
knowledge of ICT skills
management
Skills for modern Skills for creating
Dr.Sc
educational tech- and use of e-learning
PhD
nologies tools
Magister
Specialist
Bachelor
Levels 5-9
Personality Professional
Competencies Skill
Leader skills Planning
Critical thinking Design
Holistic view of a Resources management
Cooperation and collaboration
problem Performance activities
Responsibility Monitoring and self evaluation
and activity
Fig. 1. A holistic model of a specialist
442 M. Shyshkina
All the components of a specialist’s competencies, skills and knowledge are con-
sistently formed within the main level of education which corresponds to National
qualification framework (levels 5-9).
Cloud computing decision is a reasonable way to support holistic learning settings
giving a platform for unite representation and access to learning tools for different
levels and domain of education as also for different individuals and groups of users.
Promising ways of assessing resources quality, while building a holistic learning
environment are:
A. Analysis of the most appropriate ways to use cloud computing technology to
supplementing and structuring collection of educational learning resources, fill-
ing it with the resources on this basis and organizing multiple access to their
use
B. Use of a certain set of educational resources for testing methods to evaluate the
quality of their use within the cloud-based infrastructure of organization
C. Recommendations on methods to replenish the collection, its prototyping and
ways of structuring resources
D. Elaboration of requirements to provide electronic resources, for collection re-
plenishment
E. Analysis of cloud computing technology outsourcing for optimal selection and
use of resources’ collections
F. Creation of recommendations to developers and material for replenishment and
application of existing electronic learning resources
G. Development of recommendations for dissemination and use of collections of
electronic resources
6 An Expected Impact and Social Results of the Project
The important step to wider application and introduction of new learning approaches
and to gain most possible benefit from emerging technologies and ICT tools should be
achieved through modernization and upgrading of ICT learning environment of edu-
cational institutions, increasing of overall level of e-learning.
To achieve these goals the main problem is to rise ICT and professional level of
competencies of subjects of the learning process – managers, pedagogical personnel
and staff and also personal of ICT departments. Just the people are the most valuable
factor of empowerment of development and formation of social and economical sys-
tems and educational systems in particular. Just the people are the most important
resource which should be involved so as to improve the quality of these social sys-
tems and to manage their purposeful and productive growth. By this reason develop-
ment of tools and resources to train teachers and stuff is critical point because it really
concern to all levels of educational systems functioning.
The whole impact of implementation of learning tools and techniques based on
cloud computing is aimed at:
Broaden use of ICT in education aiming at wider take up by learners and teachers
Effective public-private partnerships for introduction and managements of learning
environment solutions
Holistic Approach to Training of ICT Skilled Educational Personnel 443
More efficient introduction of ICT into the learning process through the exploita-
tion of monitoring and assessment tools
More timely and purposeful acquisition of skills and competences through ICT-
based learning technologies, in educational establishments and public administra-
tions
Increased involvement with the adoption of learning digital technologies
The important step to wider application and introduction of new learning ap-
proaches and to gain most possible benefit from emerging technologies and ICT tools
should be achieved through modernization and upgrading of ICT learning environ-
ment of educational institutions, development of new learning approaches, creating
more advanced learning technologies [1], [2].
Formation of innovative ICT infrastructure of the institution could solve some of
the problems of development highly skilled educational and management personnel,
bridging the gap between the process of training and the level of demand for their
product.
Due to development of cloud computing technologies opportunities, functionality
and access to collections of electronic learning resources has significantly increased.
In this regard, cloud computing is a promising direction of development of electronic
resources’ collections (may be relevant for development of collections), as it allows
the creation of a unified methodology for a single platform, a framework for devel-
opment and testing, and for improvement and elaboration of integrated assessment
methods’ quality. This gives an added value to available recourses [2], [11].
The social results will help to modernize the learning environment of educational
institutions and organizations, to increase educational potential of ICT and add value
to the best examples of available learning resources due to their flexible and learner-
adaptive access.
At the same time there are several aspects of the cloud-based learning architecture
to be a subjected to further research. There are problems of pedagogical and psycho-
logical support in regard to the processes of the design and organization of an educa-
tional institution’s cloud infrastructure, prospecting possible organizational structures
to provide learning environment functioning and to teach educational managers and
organizers, pedagogical and technical stuff how to use new methods and approaches
to learning, based on cloud computing. There is a necessity therefore, to create an
educational and training system of support used by management personnel, teachers
and learners.
The result of instrumentation for cloud-based learning resources collection elabora-
tion, and development of cloud-based learning environment of educational institution
might be used within different learning and organizational educational structures.
7 Analysis and Estimation of Perspective Ways of Development
The cloud based learning infrastructure is to give the opportunities:
To combine the processes of development and use of electronic resources to sup-
port learner competencies
444 M. Shyshkina
To insure holistic approach to specialist education and training, combining both
technological and social competences, development of critical skills of a learner
To integrate the processes of training, retraining and advanced training, at different
levels of education by providing access to electronic resources of a unite learning
environment
To solve or significantly mitigate the problems of association of electronic re-
sources of the institution into unite framework
To access to the best examples of electronic resources and services to those units or
institutions, where there is no strong ICT support services for e-learning
To provide of invariant access to learning resources within the unified educational
environment, depending on the purpose of study or educational level of the student,
enabling person-oriented approach to learning
To make conditions for a higher level of harmonization, standardization and qual-
ity of electronic resources, which may lead to emergence of the better examples of
learning resources and to more massive use them
8 Conclusion
There are real advantages of CC technologies to assure more flexible, scalable and
cost-effective decisions of access to learning resources as within the learning envi-
ronment of the university and also in learning environment of the whole region, na-
tional and international scale. This is an advantage so as to ensure joint use and wid-
ening participation in the learning courses of learners from different institution were
necessary services are substantiated and supported. As if holistic approaches to cloud
services development are already used in education so the challenge is to transfer this
experience into wider context.
The project is implemented within the framework of the joint research laboratory
of Cloud computing in education of the Institute of Information Technologies and
Learning Tools of NAPS of Ukraine (Kiev) and the Krivoy Rog State University
(Krivoy Rog), www.ccelab.ho.ua.
References
1. Bykov, V.: Models of Organizational Systems of Open Education. Atika, Kyiv (2009) (in
Ukrainian)
2. Bykov V.: Cloud Computing Technologies, ICT Outsourcing, and New Functions of ICT
Departments of Educational and Research Institutions. Information Technologies in Edu-
cation, 10, 8–23 (2011) (in Ukrainian)
3. Bykov V., Shyshkina M.: Innovative Models of Education and Training of Skilled Person-
nel for High Tech Industries in Ukraine. Information Technologies in Education, 15, 19–
29 (2013)
4. Candis Best, K.: Holistic Leadership: a Model for Leader-Member Engagement and De-
velopment. The Journal of Values Based Leadership, 4(1) (2011)
5. Cheetham, G., Chivers, G.: Towards a Holistic Model of Professional Competence. Jour-
nal of European Industrial Training, 20(5), 20–30 (1996)
Holistic Approach to Training of ICT Skilled Educational Personnel 445
6. Chapman B., Mosher F. C., Page E. C.: Public Administration. Encyclopedia Britannica,
http://www.britannica.com/EBchecked/topic/482290/public-administration
7. Forbes, S. H., Martin, R. A.: What Holistic Education Claims About Itself: an Analysis of
Holistic Schools’ Literature. In: Proc. Annual Conf. American Education Research Asso-
ciation, San Diego, California (2004)
8. Zhang, Qi, Cheng, Lu, Boutaba, R.: Cloud Computing: State-of-the-Art and Research
Challenges. J. Internet Serv. Appl., 1, 7–18 (2010)
9. Quatro, S. A., Waldman, D. A. Galvin, B.M.: Developing Holistic Leaders: Four Domains
for Leadership Development and Practice. Human Resource Management Review, 17,
427–441 (2007)
10. Kettl, D.; Fessler J.: The Politics of the Administrative Process. CQ Press, Washington
D.C. (2009)
11. Shyshkina, M.: Innovative Technologies for Development of Learning Research Space of
Educational Institution. Information Technologies and Society, 1,
http://ifets.ieee.org/russian/depository/v16_i1/pdf/15.pdf (2013) (In Russian)
12. Sultan, N.: Cloud Computing for Education: A New Dawn? Int. J. of Information Man-
agement, 30, 109–116 (2010)
2.3 2nd International Workshop
on Algebraic, Logical, and Algorithmic
Methods of System Modeling,
Specification and Verification
(SMSV 2013)
Foreword
It is our pleasure to offer you the selection of papers for the 2nd International Work-
shop on Algebraic, Logical, and Algorithmic Methods of System Modeling, Specifi-
cation and Verification (SMSV 2013) which has been co-located with the 9-th Inter-
national Conference on ICT in Education, Research, and Industrial Applications:
Integration, Harmonization, and Knowledge Transfer (ICTERI 2013) held at Kherson,
Ukraine on June 19-22, 2013.
Workshop SMSV 2013 is a successor of the International Workshop on Algebraic,
Logical, and Algorithmic Methods of System Modeling, Specification and Verifica-
tion (SMSV 2012) which was held in Kherson (Ukraine) on June, 6-12, 2012. The
SMSV 2013 was organized by Kherson State University, Taras Shevchenko National
University of Kyiv, and Paul Sabatier University of Toulouse within the framework of
the cooperation agreement between the universities. The workshop attracted scientists
from Austria, France, Algeria, Russia, and Ukraine.
Presented papers demonstrated the interest in the topics on different formal meth-
ods of system development, and we plan to organize such workshops on a regular
basis. We hope that presentations and discussions will help to identify topics of mu-
tual interest that can be considered as a base of project proposals to be submitted to
international scientific programs.
June, 2013 Vladimir Peschanenko
Mykola Nikitchenko
Martin Strecker
An Abstract Block Formalism for Engineering
Systems?
Ievgen Ivanov1,2
1
Taras Shevchenko National University of Kyiv, Ukraine
2
Paul Sabatier University, Toulouse, France
ivanov.eugen@gmail.com
Abstract. We propose an abstract block diagram formalism based on
the notions of a signal as a time-varying quantity, a block as a signal
transformer, a connection between blocks as a signal equality constraint,
and a block diagram as a collection of interconnected blocks. It does not
enforce implementation details (like internal state-space) or particular
kinds of dynamic behavior (like alternation of discrete steps and contin-
uous evolutions) on blocks and can be considered as an abstraction of
block diagram languages used by engineering system designers.
We study its properties and give general conditions for well-definedness of
the operation of a system specified by a block diagram for each admissible
input signal(s).
Keywords. block diagram, signal transformer, semantics, engineering
system
Key Terms. Mathematical Model, Specification Process, Verification
Process
1 Introduction
Many software tools for developing control, signal processing, and communica-
tion systems are based on block diagram notations familiar to control engineers.
Examples include system design software Simulink [1], Scicos [2], Dymola [3],
SCADE [4], declarative synchronous languages [5], some embedded program-
ming languages [6, 7].
In such notations, a diagram consists of blocks (components) connected by
links. Typically, blocks have input and output ports, and (directed) links connect
output ports of one block with input ports of the same or another block. A block
is interpreted as an operation which transforms input signals (i.e. time-varying
quantities) flowing though its input ports into output signals.
?
Part of this research has been supported by the project Verisync (ANR-10-BLAN-
0310), France.
An Abstract Block Formalism for Engineering Systems 449
Wide applicability of block diagram notations makes them an interesting
object of study from a theoretical perspective. Classical control theory and signal
processing already provide some degree of formal treatment of block diagrams
[8, 9], but this is normally not sufficient to handle such aspects of modern system
design languages as mixing of analog and discrete-time blocks, partially defined
block operations, non-numeric data processing, etc.
To take these issues into account, researches developed formal semantics for
various block diagram languages [10–13]. Some effort has been made to unify
approaches taken by different engineering system modeling and analysis tools
and make them interoperable by the use of exchange languages with well-defined
semantics [14, 15] such as Hybrid System Interchange Format (HSIF) which gives
semantics of hybrids systems in terms of dynamic networks of hybrid automata
[16, 17].
Although hybrid automata-based approaches like HSIF can be used to give
semantics to block diagram languages [18] and have many advantages (e.g. avail-
ability of verification theory for hybrid automata), we consider them not entirely
satisfactory from a theoretical standpoint for the following reasons:
– Semantics of a system component (block) is based on the notion of a dy-
namical system with an internal state, and so the components with the same
externally observable behavior can be semantically distinguishable. In our
opinion, this is not needed for a high-level semantics which does not intend
to describe details of the component’s physical/logical implementation.
– Semantics of a system has a computational nature: it describes a sequence
of discrete steps, where a step may involve function computations, solving
initial-value problems for differential equations (continuous evolution), etc.
It may be adequate for certain classes of discrete-continuous systems, but it
does not always capture the behavior of a physical realization of a system
(and thus may conflict which the view of a system designer).
For example, a Zeno execution [17] of a hybrid automaton can be described
as an infinite sequence of discrete steps which takes a bounded total time
(but each step takes a non-zero time). This normally does not correspond
to the behavior of a physical system described by the automaton. In many
cases this is caused by system modeling simplifications. The conflict is usu-
ally resolved by applying a certain method of continuation of an execution
beyond Zeno time (regularization, Fillipov solution, etc.) [19]. But an ex-
tended execution is not, in fact, a sequence of discrete steps, as it resumes
after the accumulation point.
In our opinion, in the general case, the dynamic behavior of a system should
not be restricted to a particular scheme like a sequence of discrete steps and
continuous evolutions.
The goal of this paper is to introduce abstract formal models for blocks
and block diagrams which overcomes limitations of the classical control/signal-
theoretic approach to them and does not enforce implementation details (like
internal state-space) or particular kinds of dynamic behavior (like alternation of
discrete steps and continuous evolutions) on blocks.
450 Ie. Ivanov
These models can be used to identify the most general properties of block
diagram languages which are valid regardless of implementation details. In par-
ticular, in the paper we will give a general formulation and conditions for well-
definedness of the operation of a system specified by a block diagram for each
admissible input signal(s).
To achieve our goal, we will use a composition-nominative approach [20].
The main idea of this approach is that semantics of a system is constructed
from semantics of components using special operations called compositions, and
the syntactic representation of a system reflects this construction.
The paper is organized in the following way:
– In Section 2 we give definitions of the auxiliary notions which are used in
the rest of the paper.
– In Section 3 we introduce abstract notions of a block, a connection, a block
diagram, and a block composition. We show how they fit into a system design
process and give conditions of well-definedness of the operation of a system
specified by a block diagram for each admissible input signal(s).
2 Preliminaries
2.1 Notation
We will use the following notation: N = {1, 2, 3, ...}, N0 = N ∪ {0}, R+ is the
set of nonnegative real numbers, f : A → B is a total function from A to B,
˜ is a partial function from A to B, 2A is the power set of a set A, f |X is
f : A→B
the restriction of a function f to a set X. If A, B are sets, then B A denotes the
set of all total functions from A to B. For a function f : A→B˜ the symbol f (x) ↓
(f (x) ↑) means that f (x) is defined (respectively undefined) on the argument x.
We denote the domain and range of a function as dom(f ) = {x | f (x) ↓} and
range(f ) = {y | ∃x f (x) ↓ ∧ y = f (x)} respectively. We will use the the same
notation for the domain and range of a binary relation: if R ⊆ A × B, then
dom(R) = {x | ∃ y (x, y) ∈ R} and range(R) = {y | ∃ x (x, y) ∈ R}.
We will use the notation f (x) ∼ = g(x) for the strong equality (where f and g
are partial functions): f (x) ↓ iff g(x) ↓ and f (x) ↓ implies f (x) = g(x).
The symbol ◦ denotes a functional composition: (f ◦ g)(x) ∼ = g(f (x)).
By T we denote the (positive real) time scale [0, +∞). We assume that T is
equipped with a topology induced by the standard topology on R.
Additionally, we define the following class of sets:
T0 = {∅, T } ∪ {[0, x) | x ∈ T \{0}} ∪ {[0, x] | x ∈ T }
i.e. the set of (possibly empty, bounded or unbounded) intervals with left end 0.
2.2 Multi-valued functions
A multi-valued function [20] assigns one or more resulting values to each argu-
ment value. An application of a multi-valued function to an argument is inter-
preted as a nondeterministic choice of a result.
An Abstract Block Formalism for Engineering Systems 451
Definition 1 ([20]). A (total) multi-valued function from a set A to a set B
tm
(denoted as f : A−→ B) is a function f : A → 2B \{∅}.
Thus the inclusion y ∈ f (x) means that y is a possible value of f on x.
2.3 Named sets
We will use a simple notion of a named set to formalize an assignment of values
to variable names in program and system semantics.
Definition 2. ([20]) A named set is a partial function f : V →W
˜ from a non-
empty set of names V to a set of values W .
In this definition both names and values are unstructured. A named set can be
considered as a partial (”flat”) case of a more general notion of nominative data
[20] which reflects hierarchical data organizations and naming schemes.
We will use a special notation for the set of named sets: V W denotes the set
of all named sets f : V →W
˜ (this notation just emphasises that V is interpreted
as a set of names). We consider named sets equal, if their graphs are equal.
An expression of the form [n1 7→ a1 , n2 7→ a2 , ...] (where n1 , n2 , ... are distinct
names) denotes a named set d such that the graph of d is {(n1 , a1 ), (n2 , a2 ), ...}.
A nowhere-defined named set is called an empty named set and is denoted as [].
For any named sets d1 , d2 we write d1 ⊆ d2 (named set inclusion), if the
graph of a function d1 is a subset of the graph of d2 . We extend set-theoretical
operations of union ∪, intersection ∩ and difference \ to the partial operations on
named sets in the following way: the result of a union (intersection, difference) of
named sets (operation’s arguments) is a named set d such that the graph of d is
the union (intersection, difference) of graphs of the arguments (if such d exists).
3 An Abstract Block Formalism
3.1 Abstract block
Let us introduce abstract notions of a signal as a time-varying quantity and
a block as a signal transformer. We will use a real time scale for signals, but
we will not require them to be continuous or real-valued. So the signals can be
piecewise-constant as well and can be used to represent discrete evolutions.
Informally, a block (see Fig. 1) is a device which receives input signals and
produces output signals. We call a collection of input (output) signals an input
(resp. output) signal bunch. At each time moment (the value of) a given signal
may be present or absent. In the general case, the presence of an input signal at
a given time does not imply the presence of an output signal at the same or any
other time moment.
A block can operate nondeterministically, i.e. for one input signal bunch it
may choose an output signal bunch from a set of possible variants. However,
for any input signal bunch there exists at least one corresponding output signal
452 Ie. Ivanov
bunch (although the values of all signals in it may be absent at all times, which
means that the block does not produce any output values).
Normally, a block processes the whole input signal bunch, and does or does
not produce output values. However, in certain cases a block may not process
the whole input signal bunch and may terminate at some time moment before
its end. This situation is interpreted as an abnormal termination of a block (e.g.
caused by an invalid input).
Fig. 1. An illustration of a block with input signals x1 , x2 and output signals y1 ,
y2 . The plot displays example evolutions of input and output signals. The input and
output signals are lumped into an input and output signal bunch respectively. Solid
curves represent (present) signal values. Dashed horizonal segments indicate absence
of a signal value. Dashed vertical lines indicate the right boundaries of the domains of
signal bunches.
Let us give formal definitions. Let W be a (fixed) non-empty set of values.
Definition 3. (1) A signal is a partial function from T to W (f : T →W ˜ ).
˜ V W such
(2) A V -signal bunch (where V is a set of names) is a function s : T →
that dom(s) ∈ T0 . The set of all V -signal bunches is denoted as Sb(V, W ).
(3) A signal bunch is a V -signal bunch for some V .
(4) A signal bunch s is trivial, if dom(s) = ∅ and is total, if dom(s) = T . A
trivial signal bunch is denoted as ⊥.
An Abstract Block Formalism for Engineering Systems 453
(5) For a given signal bunch s, a signal corresponding to a name x is a partial
function t 7→ s(t)(x). This signal is denoted as s[x].
(6) A signal bunch s1 is a prefix of a signal bunch s2 (denoted as s1 s2 ), if
s1 = s2 |A for some A ∈ T0 .
Note that on V -signal bunches is a partial order (for an arbitrary V ). Later we
will need generalized versions of the prefix relation for pairs and indexed families
of pairs of signal bunches.
For any signal bunches s1 , s2 , s01 , s02 let us denote (s1 , s2 ) 2 (s01 , s02 ) iff there
exists A ∈ T0 such that s1 = s01 |A and s2 = s02 |A .
For any indexed families of pairs of signal bunches (sj , s0j )j∈J and (s00j , s000 j )j∈J
0 J,2 00 000
of signal bunches let us denote (sj , sj )j∈J (sj , sj )j∈J iff there exists A ∈ T0
such that sj = s00j |A and s0j = s000
j |A for all j ∈ J.
It is easy to check that 2 is a partial order on pairs of signal bunches and
J,2 is a partial order on J-indexed families of pairs of signal bunches.
A block has a syntactic aspect (e.g. a description in a specification language)
and a semantic aspect – a partial multi-valued function on signal bunches.
Definition 4. (1) A block is an object B (syntactic aspect) together with an as-
sociated set of input names In(B), a set of output names Out(B), and a total
tm
multi-valued function Op(B) : Sb(In(B), W )−→ Sb(Out(B), W ) (operation,
semantic aspect) such that o ∈ Op(B)(i) implies dom(o) ⊆ dom(i).
(2) Two blocks B1 , B2 are semantically identical, if In(B1 ) = In(B2 ), Out(B1 ) =
Out(B2 ), and Op(B1 ) = Op(B2 ).
(3) An I/O pair of a block B is a pair of signal bunches (i, o) such that o ∈
Op(B)(i). The set of all I/O pairs of B is denoted as IO(B) and is called
the input-output (I/O) relation of B.
An inclusion o ∈ Op(B)(i) means that o is a possible output of a block B on the
input i. For each input i there is some output o. The domain of o is a subset of
the domain of i. If o becomes undefined at some time t, but i is still defined at
t, we interpret this as an error during the operation of the block B (the block
cannot resume its operation after t).
Definition 5. A block B is deterministic, if Op(B)(i) is a singleton set for each
In(B)-signal bunch i.
We interpret the operation of a block as a (possibly nondeterministic) choice
of an output signal bunch corresponding to a given input signal bunch. However,
we would also like to describe this choice as dynamic, i.e. that a block chooses
the output signal values at each time t, and in doing so it cannot rely on the
future values of the input signals (i.e. values of the input signals at times t0 > t).
If a block is deterministic, this requirement can be formalized in the same
way as the notion of a causal (or nonanticipative) input-output system [26].
Definition 6. A deterministic block B is causal iff for all signal bunches i1 , i2
and A ∈ T0 , o1 ∈ Op(B)(i1 ), o2 ∈ Op(B)(i2 ), the equality i1 |A = i2 |A implies
o1 | A = o 2 | A .
454 Ie. Ivanov
This means that the value of the output signal bunch at time t can depend only
on the values of the input signal at times ≤ t.
Some works in the domain of systems theory extend the notion of a causal
(deterministic) system to nondeterministic systems. However, there is no unified
approach to an extension of this kind. For example, in the work [21], a system,
considered as a binary relation on (total) signals S ⊆ AT × B T , where T is
a time domain (Mesarovic time system, [22]) is “non-anticipatory”, if itSis a
union of (graphs of) causal (non-anticipatory) selections from S, i.e, S = {f :
dom(S) → range(S) | f ⊆ S, f is causal}. In the work [23] the authors define
another notion of a “non-anticipatory” or “causal” system in nondeterministic
case. In the theory developed in the work [24], the authors use a similar notion
of a “precausal” system, which is also defined in [22], as a generalization of the
notion of a causal system to the nondeterministic case.
In this work, we generalize the notion of a non-anticipatory system in sense
of [23] to blocks and call such blocks nonanticipative, and generalize the notion
of a non-anticipatory system in sense of [21] to blocks, but call such blocks
strongly nonanticipative. We will show that strongly nonanticipative block is
nonanticipative. We will consider the words “causal” and “nonanticipative” as
synonyms when they are used informally, but we will distinguish them in the
context of formal definitions to avoid a conflict with Definition 6.
Note, however, that the notion of a strongly nonanticipative block defined
below is very different from the notion of a ”strictly causal“ system, defined in
some works [25] as a system which uses only past (but not current or future)
values of the input signal(s) to produce a current value of the output signal(s).
Definition 7. A block B is nonanticipative, if for each A ∈ T0 and i1 , i2 ∈
Sb(In(B), W ), if i1 |A = i2 |A , then
{o|A | o ∈ Op(B)(i1 )} = {o|A | o ∈ Op(B)(i2 )}.
Definition 8. A block B is a sub-block of a block B 0 (denoted as B E B 0 ), if
In(B) = In(B 0 ), Out(B) = Out(B 0 ), and IO(B) ⊆ IO(B 0 ).
Informally, a sub-block narrows nondeterminism of a block.
Definition 9. A block B is strongly nonanticipative, if for each (i, o) ∈ IO(B)
there exists a deterministic causal sub-block B 0 E B such that (i, o) ∈ IO(B 0 ).
Informally, the operation of a strongly nonanticipative block B can be inter-
preted as a two-step process:
1. before receiving the input signals, the block B (nondeterministically) chooses
a deterministic causal sub-block B 0 E B (response strategy);
2. the block B 0 receives input signals of B and produces the corresponding
output signals (response) which become the output signals of B.
Intuitively, it is clear that in this scheme at any time the block B does not need
a knowledge of the future of its input signals in order produce the corresponding
output signals.
An Abstract Block Formalism for Engineering Systems 455
Lemma 1. If B is a deterministic block, then B is causal iff B is nonanticipa-
tive.
Proof. Follows immediately from Definition 6.
The following theorem gives a characterization of a nonanticipative block
which does not rely on comparison of sets of signal bunches.
Theorem 1. A block B is nonanticipative iff the following holds:
(1) if (i, o) ∈ IO(B) and (i0 , o0 ) 2 (i, o), then (i0 , o0 ) ∈ IO(B);
(2) if o ∈ Op(B)(i) and i i0 , then (i, o) 2 (i0 , o0 ) for some o0 ∈ Op(B)(i0 ).
Proof. (1) Assume that (1) and (2) are satisfied. Assume that A ∈ T0 , i1 , i2 ∈
Sb(In(B), W ), and i1 |A = i2 |A . Let o ∈ Op(B)(i1 ). Then from assump-
tion (1) we have o|A ∈ Op(B)(i1 |A ), because (i1 |A , o|A ) 2 (i1 , o). More-
over, i1 |A i2 , because i1 |A = i2 |A . Thus (i1 |A , o|A ) 2 (i2 , o0 ) for some
o0 ∈ Op(B)(i2 ) by assumption (2). It is not difficult to check that o|A ∈
{o00 |A | o00 ∈ Op(B)(i2 )}. Because i1 , i2 , A are arbitrary, B is nonanticipative
by Definition 7.
(2) Assume that B is nonanticipative. Let us prove (1). Assume that (i, o) ∈
IO(B) and (i0 , o0 ) 2 (i, o). Then i0 = i|A and o0 = o|A for some A ∈ T0 .
Then i0 |A = (i|A )|A = i|A , whence
o0 = o|A ∈ {o00 |A | o00 ∈ Op(B)(i)} = {o00 |A | o00 ∈ Op(B)(i0 )}
by Definition 7. Then o0 = o00 |A for some o00 ∈ Op(B)(i0 ). Moreover, dom(o00 ) ⊆
dom(i0 ) ⊆ A. Thus o0 = o00 and (i0 , o0 ) ∈ IO(B).
Let us prove (2). Assume that o ∈ Op(B)(i) and i i0 . Then i = i0 |A for
some A ∈ T0 . Then i|A = (i0 |A )|A = i0 |A , whence
o|A ∈ {o00 |A | o00 ∈ Op(B)(i)} = {o00 |A | o00 ∈ Op(B)(i0 )}
by Definition 7. Then o|A = o0 |A for some o0 ∈ Op(B)(i0 ). Moveover, dom(o) ⊆
dom(i) ⊆ A, whence o = o|A = o0 |A . Thus (i, o) 2 (i0 , o0 ).
Theorem 2. (About strongly nonanticipative block)
(1) If a block B is strongly nonanticipative, then it is nonanticipative.
(2) There exists a nonanticipative block which is not strongly nonanticipative.
Proof (Sketch).
(1) Assume that B is strongly nonanticipative. Let R be the set of all re-
lations R ⊆ IO(B) such that R is an I/O relation of a nonanticipative
block. For each R ∈ IO let us define a block BR such that IO(BR ) = R,
In(BR ) = In(B), Out(BR ) = Out(B). Let B = {BR | R ∈ R}. Then each
element ofSB is nonanticipative.
S From Definition 9 and Lemma 1 we have
0
IO(B) ⊆ R = B 0 ∈B IO(B S ). On the other hand, IO(B 0 ) ⊆ IO(B) for
any B ∈ R, so IO(B) = B 0 ∈B IO(B 0 ). It is easy to see from Theorem 1
0
that (nonempty) union of I/O relations of nonanticipative block is an I/O
relation of a nonanticipative block. Thus B is nonanticipative.
456 Ie. Ivanov
(2) Assume that W = R. Let f : R → R be a function that is discontinuous
at some point (e.g. a signum function). Let use define a block B such that
In(B) = {x} and Out(B) = {y} for some names x, y, and for each i ∈
Sb(In(B), R), Op(B)(i) is defined as follows:
• if dom(i[x]) = T and limt→+∞ i[x](t) exists and finite, then Op(B)(i) is
the set of all {y}-signal bunches o such that dom(o) = dom(o[y]) = T
and
lim o[y](t) = f lim i[x](t) ;
t→+∞ t→+∞
• otherwise, Op(B)(i) is S the set of all {y}-signal bunches o such that
dom(o) = dom(o[y]) = {A ∈ T0 | A ⊆ dom(i[x])}.
Obviously, in this definition Op(B)(i) 6= ∅ (because T0 is closed under
unions) and dom(o) ⊆ dom(i) for each o ∈ Op(B)(i). So B is indeed a
block. Using Theorem 1 it is not difficult to show that B is nonanticipa-
tive. Suppose that B has a a deterministic causal sub-block B 0 . Let a ∈ R
and ak ∈ R, k = 1, 2, ... be a sequence such that limk→∞ ak = a. Let us
show that limk→∞ f (ak ) = f (a). Let us define sequences ik ∈ Sb({y}, R),
ok ∈ Sb({y}, R), and tk ∈ T , k = 1, 2, ... by induction as follows.
Let i1 (t) = [x 7→ a1 ] for all t ∈ T , o1 be a unique member of Op(B 0 )(i1 ),
and t1 = 0. If i1 , i2 , ..., ik are already defined, let ik+1 (t) = ik (t), if t ∈ [0, tk ]
and ik+1 (t) = [x 7→ ak+1 ], if t ∈ T \[0, tk ]. Let ok+1 be a unique member
of Op(B 0 )(ik+1 ). Because B 0 E B, dom(ok+1 ) = dom(ok+1 [y]) = T and
limt→+∞ ok+1 [y](t) = f (limt→+∞ ik+1 [x](t)) = f (ak+1 ). Then let
tk+1 = 1 + max{tk , inf{τ ∈ T |
1
sup{|ok+1 [y](t) − f (ak+1 )| | t ≥ τ } ≤ }}
k+1
We have defined sequences ik , ok , tk . The sequence tk , k = 1, 2, ... is a strictly
increasing and unbounded from above and t1 = 0.
Let i be a {x}-signal bunch such that dom(i) = T , i(t1 ) = i1 (t1 ), and
i(t) = ik+1 (t), if t ∈ (tk , tk+1 ], k ∈ N, and o be a (unique) member of
Op(B 0 )(i). We have ik+1 [x](t) = ak+1 for all k = 1, 2, ... and t > tk . Then
i[x](t) ∈ {ak+1 , ak+2 , ...} for all k ∈ N and t > tk . For each > 0 there exists
k ∈ N such that |ak0 − a| < for all k 0 ≥ k, whence |i[x](t) − a| < for
all t > tk . Thus limt→+∞ i[x](t) = a. Then dom(o) = dom(o[y]) = T and
limt→+∞ o[y](t) = f (a), because B 0 E B.
On the other hand, ik+1 |[0,tk ] = ik |[0,tk ] for all k ∈ N. Because tk is an
increasing sequence, we have ik0 |[0,tk ] = ik |[0,tk ] for all k and k 0 ≥ k. Be-
sides, i|(tk ,tk+1 ] = ik+1 |(tk ,tk+1 ] for all k ∈ N, whence i|(tk ,tk+1 ] = ik0 |(tk ,tk+1 ]
for all k 0 ≥ k + 1. Also, ik (t1 ) = i1 (t1 ) for all k ∈ N. Then i|[0,tk ] =
i|{t1 }∪(t1 ,t2 ]∪...∪(tk−1 ,tk ] = ik |[0,tk ] for all k = 2, 3, ..., whence o|[0,tk ] = ok |[0,tk ] ,
because B 0 is causal. Then o(tk ) = ok (tk ) for all k = 2, 3, ..., and from the
definition of tk we have |o[y](tk ) − f (ak )| = |ok [y](tk ) − f (ak )| ≤ k1 for all k =
2, 3, .... This implies that limk→∞ f (ak ) = f (a), because limt→+∞ o[y](t) =
f (a). We conclude that f is sequentially continuous and thus is continuous.
An Abstract Block Formalism for Engineering Systems 457
This contradicts our choice of f as a discontinuous function. Thus B has not
deterministic causal sub-blocks. Consequently, B is not strongly nonantici-
pative, though it is nonanticipative.
The proof of this theorem gives a reason of why Definition 9 better captures
a intuitive idea of causality than Definition 7. Consider, for example, the block
B constructed in the proof of the item (2) of Theorem 2, when f is the signum
function (i.e., f (0) = 0, f (x) = 1, if x > 0, and f (x) < 0, if x < 0). Then B
outputs a signal which converges to 1 (as t → +∞) whenever the input signal
converges to a positive number (as t → +∞). Moreover, it outputs a signal which
converges to 0 whenever the input signal converges to 0. This implies that when
the block receives a decreasing positive input signal which tends to 0, it decides
to output values which are close to 0 starting from some time t. Intuitively, after
reading the input signal until time t, the block decides that 0 is a more likely
limit of the input signal than a positive value, but such a decision cannot be
based on the past values of the input signal, so it requires some knowledge of
the future of the input signal. These informal observations are captured by the
fact that B has no deterministic causal sub-blocks.
In the rest of the paper we will focus on strongly nonanticipative blocks, as
more adequate models of (real-time) information processing systems.
Consider an example of a strongly nonanticipative block.
Let u, y be names. Assume that W = R.
Example 1. Let B be a block such that In(B) = {u}, Out(B) = {y}, and for
each i, Op(B)(i) = {o1 (i), o2 (i)}, where o1 (i), o2 (i) ∈ Sb(Out(B), W ) are signal
bunches such that
– dom(o1 (i)) = dom(o2 (i)) = dom(i);
– o1 (i)(t) = [y 7→ i[u](t)] for all t ∈ dom(i);
– o2 (i)(t) = [y 7→ 2i[u](t)] for all t ∈ dom(i).
Informally, this means that B is a gain block with a slope which is either 1 or 2
during the whole duration of the block’s operation.
Obviously, B satisfies Definition 4(1). Let us check that it is strongly nonan-
ticipative. For j = 1, 2 let Bj Θ B be a sub-block such that Op(Bj )(i) = {oj (i)}
for all i ∈ Sb(In(B), W ) (i.e. B1 always selects o1 (i) from Op(B)(i) and B2
always selects o2 (i)).
The blocks B1 , B2 are deterministic and it is easy to see that they are causal.
Obviously, each I/O pair (i, o) ∈ IO(B) belongs either to IO(B1 ), or to IO(B2 ),
so B is strongly nonanticipative.
Now let us consider an example of a block which is not nonanticipative.
Example 2. Let B 0 be a block such that In(B 0 ) = {u}, Out(B 0 ) = {y}, and
– Op(B 0 )(i) = {o1 }, where dom(o1 ) = dom(i) and o1 (t) = [y 7→ 1] for all
t ∈ dom(i), if dom(i[u]) = T ;
458 Ie. Ivanov
– Op(B 0 )(i) = {o2 }, where dom(o2 ) = dom(i) and o2 (t) = [y 7→ 0] for all
t ∈ dom(i), otherwise.
Informally, the block B 0 decides whether its input signal u is total. It is easy to
see that B 0 indeed satisfies Definition 4(1), but the condition (1) of Theorem 1 is
not satisfied, because (i, o) ∈ IO(B 0 ), where i(t) = [u 7→ 0] for all t ∈ T , o(t) =
/ IO(B 0 ).
[y 7→ 1] for all t ∈ T , and (i|[0,1] , o|[0,1] ) 2 (i, o), but (i|[0,1] , o|[0,1] ) ∈
0
So B is not nonanticipative. Informally, the reason is that at each time t the
current value of y depends on the entire input signal.
3.2 Composition of blocks
By connecting inputs and outputs of several (strongly nonanticipative) blocks
one can form a larger block – a composition of blocks (see Fig. 2). We assume that
an output can be connected to several inputs, but each input can be connected
to no more than one output. Unconnected inputs and outputs of constituent
blocks become inputs and output of the composition. Connections are interpreted
as signal equality constraints and they always relate an output of some block
(”source”) with an input of the same or another block (”target”). We represent
connections in the graphical form (like in Fig. 2) as arrows connecting blocks.
Fig. 2. An informal illustration of a block composition. Three blocks are composed
to form a larger block (a dashed rectangle). Solid arrows denote connections between
blocks. Dashed arrows denote unconnected inputs/outputs of the blocks 1, 2, 3 (which
become the input/outputs of the dashed block).
Definition 10. (1) A block diagram is a pair ((Bj )j∈J , ) of an indexed family
of blocks (Bj )j∈J and an injective binary relation ⊆ Vout × Vin , which is
called an interconnection relation, where
[ [
Vin = {j} × In(Bj ), Vout = {j} × Out(Bj ).
j∈J j∈J
An Abstract Block Formalism for Engineering Systems 459
Fig. 3. A strongly nonanticipative block B is composed to obtain a block B 0 (a dashed
rectangle). A solid loop arrow denotes a connection between an output and an input
of B. Dashed arrows denote unconnected inputs and outputs of B which become the
inputs and outputs of B 0 .
(2) A block diagram ((Bj )j∈J , ) is called regular, if each Bj , j ∈ J is strongly
nonanticipative.
Note that a diagram may consist of an infinite set of blocks. A relation (j, x)
(j 0 , x0 ) means that the output x of the j-th block is connected to the input x0 of
the j 0 -th block.
A block diagram is only a syntactic aspect of a block composition. We define
semantics of a block composition only for strongly nonanticipative blocks.
To describe it informally consider Fig. 3. The connection between y2 and x2
means a signal equality constraint. The block B chooses (nondeterministically)
some deterministic sub-block B0 E B. When a signal starts flowing into the
input x1, the block B 0 tries to choose the initial values for the signals of y1,
y2, x2 so that they satisfy the operation of the block B0 (Op(B0 )) and the
signals of x2 and y2 have the same values. If such initial values do not exist, the
block B 0 terminates (the output signal bunch is nowhere defined). Otherwise,
B 0 continues to operate in the similar way until either the signals of y1, y2, x2
cannot be continued, or the input signal (x1) ends.
Definition 11. Let ((Bj )j∈J , ) be a regular block diagram.
A block B is a composition of (Bj )j∈J under the interconnection relation , if
S
– In(B) = ( S j∈J {j} × In(Bj ))\range(),
– Out(B) = ( j∈J {j} × Out(Bj ))\dom(),
– Op(B)(i) is the set of all Out(B)-signal bunches o such that there exist
deterministic causal sub-blocks Bj0 E Bj , j ∈ J and an indexed family
(ij , oj )j∈J ∈ Xm (i) such that
(1) dom(o) = dom(oj ) for all j ∈ J,
(2) o[(j, x)] = oj [x] for all (j, x) ∈ Out(B),
where Xm (i) is the set of J,2 -maximal elements of X(i), and X(i) is the
set of all indexed families of pairs of signal bunches u = (ij , oj )j∈J such that
(3) dom(ij ) = dom(oj ) = dom(ij 0 ) = dom(oj 0 ) ⊆ dom(i) for all j, j 0 ∈ J,
(4) ij [x] = i|dom(ij ) [(j, x)] for each (j, x) ∈ In(B),
(5) (ij , oj ) ∈ IO(Bj0 ) for each j ∈ J,
460 Ie. Ivanov
(6) (j, x) (j 0 , x0 ) implies oj [x] = ij 0 [x0 ].
In this definition, ij and oj denote the input and output signal bunches of the
j-th block. The set Xm (i) contains maximally extended (in sense of the relation
J,2 ) indexed families of signal bunches defined on a subset of the domain of i
(the input signal bunch of B) which satisfy constraints imposed by the intercon-
nection relation. Any such family gives a possible output of B for the given i by
the condition (2), i.e. output signals of B are obtained from the output signals
of the sub-blocks Bj0 .
It is clear that any two compositions of (Bj )j∈J under are semantically
identical.
Lemma 2. (Continuity of the operation of a causal deterministic causal block).
Let B be a deterministic causal block. Let c ⊆ Sb(In(B), W ) be a non-empty
-chain, i∗ beSits supremum (in sense of ), and o∗ ∈ Op(B)(i∗ ). Then o∗ is a
supremum of i∈c Op(B)(i) (in sense of ).
Proof. Follows from Definition 6.
Theorem 3. Let ((Bj )j∈J , ) be a regular block diagram. Then
(1) A composition of (Bj )j∈J under exists.
(2) If B is a composition of (Bj )j∈J under , then B is strongly nonanticipa-
tive.
The proof follows from Lemma 2 and Definition 11.
3.3 Specification and implementation
Above we have considered a block as an abstract model of a real component.
However, it can also be considered as a specification of requirements for a compo-
nent. Let B spec , B impl be two strongly nonanticipative blocks. Let us call them
a specification block and implementation block respectively.
Definition 12. B impl is a refinement of B spec , if B impl is a sub-block of B spec .
I.e. an implementation should have the the same input and output names as a
specification, and for each input, an output of an implementation should be one
of the possible outputs of a specification. We generalize this to diagrams.
Let D = ((Bj )j∈J , ) and D0 = ((Bj0 )j∈J 0 , 0 ) be regular block diagrams.
Definition 13. D is a refinement of D0 , if J = J 0 , Bj is a refinement of Bj0
for each j ∈ J, and the relations and 0 coincide.
Theorem 4. (Compositional refinement) Let B be a composition of (Bj )j∈J
under and B 0 be a composition of (Bj0 )j∈J 0 under 0 . If D is a refinement
of D0 , then B is a refinement of B 0 .
The proof follows from Definition 9, 11, and transitivity of the sub-block relation.
This theorem can be considered as a foundation of a modular approach [29,
30] to system design:
An Abstract Block Formalism for Engineering Systems 461
1. Create specifications of the system components (Bj0 , j ∈ J 0 ) and connect
them (0 ), as if they were real components.
2. Analyze a composition of specifications (B 0 ) to ensure that any of its imple-
mentations (B 00 E B 0 ) satisfies requirements to the final system.
3. Create an implementation (Bj ) for each specification (Bj0 ).
4. Connect implementations (according to 0 ). Then the composition of im-
plementations (B) is a final system which satisfies design requirements.
We consider the steps 1 and 3 domain- and application-specific. The conclusion
of the step 4 is addressed in Theorem 4. Step 2 requires some verification method
which depends on the nature of requirements.
One of the most basic and common requirements is that the operation of
B 0 is defined on all input signal bunches which are possible in the context of a
specific application of this composition. This trivially holds because of Theorem
3 and our definition of a block. However, B 0 may terminate abnormally on some
or all input signal bunches of interest (as we have noted, we interpret the situ-
ation when o ∈ Op(B̃)(i) and dom(o) ⊂ dom(i) for some block B̃ as abnormal
termination of B̃ on i). So the requirement can be reformulated as follows: B 0
never terminates abnormally on any input signal bunch from a given set IN
(this implies that the same property holds for B). We will call this property as
well-definedness of the operation of B 0 on IN and study it in the next subsection.
3.4 Well-definedness of the operation of a composition of blocks
Let B be a block and IN be some set of In(B)-signal bunches.
Definition 14. The operation of B is well-defined on IN , if dom(i) = dom(o)
for each i ∈ IN and o ∈ Op(B)(i).
Let D = ((Bj )j∈J , ) be a regular block diagram and B be a composition of
(Bj )j∈J under . Let F be the set of all families of blocks of the form (Bj0 )j∈J ,
where for each j ∈ J, Bj0 is a deterministic causal sub-block of Bj .
For each In(B)-signal bunch i and a family of blocks F = (Bj0 )j∈J let X F (i)
be the set of all indexed families of pairs of signal bunches u = (ij , oj )j∈J which
satisfy conditions (3)-(6) of Definition 11 (for B and (Bj0 )j∈J ).
F
Let Xm (i) denote the set of all J,2 -maximal elements of X F (i). For any
indexed family of signal bunches u = (ij , oj )j∈J let O(u) denote the set of all
Out(B)-signal bunches o which satisfy conditions (1)-(2) of Definition 11.
For any u = (ij , oj )j∈J ∈ X F (i), the domains of ij , oj for all j ∈ J coincide.
Denote by cdom(u) this common domain (we assume cdom(u) = T , if J = ∅).
S S
From Definition 11 we have Op(B)(i) = F ∈F u∈X F (i) O(u) for each i.
m
Then from Definition 14 we get the following simple criterion:
Theorem 5. The operation of B is well-defined on IN iff for each i ∈ IN ,
F ∈ F , and u ∈ X F (i), if cdom(u) ⊂ dom(i), then u ∈ F
/ Xm (i).
462 Ie. Ivanov
This criterion means that B is well-defined, if each u ∈ X F (i), the common
domain of which does not cover dom(i), is extendable to a larger u0 ∈ X F (i) (in
sense of J,2 ). We will call it a local extensibility criterion, because, basically,
to prove well-definedness, we only need to show that the members of the family
u can be continued onto a time segment [0, sup cdom(u) + ] for some small
> 0 (under constraints imposed by the interconnection relation ). Locality is
especially useful when a block diagram contains ”delay” blocks (possibly working
as variable delays), because constraints imposed by connections between blocks
reduce over small time intervals.
A drawback of this criterion is that it requires checking local extensibility
of signal bunches satisfying the I/O relations (IO(Bj0 )) of arbitrarily chosen
deterministic causal sub-blocks Bj0 E Bj (condition (5) of Definition 11), which
are not explicitly expressed in terms of I/O relations of Bj , j ∈ J.
For this reason, we seek for a condition of well-definedness in terms of I/O
relations of the constituents of the composition (IO(Bj ), j ∈ J).
Let X(i) denote the set X F (i), where F = (Bj )j∈J . Note that F may not be
a member of F .
Theorem 6. The operation of B is well-defined on IN iff for each i ∈ IN and
u ∈ X(i) there exists u0 ∈ X(i) such that u J,2 u0 and cdom(u0 ) = dom(i).
The proof follows from Theorem 5 and Definition 9.
References
1. Simulink - Simulation and Model-Based Design, http://www.mathworks.com/
products/simulink
2. Campbell, S.L., Chancelier, J.-P., Nikoukhah, R.: Modeling and Simulation in
Scilab/Scicos with ScicosLab 4.4. Springer (2010)
3. Multi-Engineering Modeling and Simulation – Dymola, http://www.3ds.com/
products/catia/portfolio/dymola
4. SCADE Suite, http://www.esterel-technologies.com/products/scade-suite
5. Caspi, P., Pilaud, D., Halbwachs, N., Plaice, J.: LUSTRE: A declarative language
for programming synchronous systems. In: 14th Annual ACM Symp. on Principles
of Programming Languages, Munich, Germany, pp. 178-188 (1987)
6. Henzinger, T., Horowitz, B., Kirsch, C.: Giotto: A Time-Triggered Language for
Embedded Programming. First International Workshop on Embedded Software,
EMSOFT’01, pp. 166-184 (2001)
7. Lublinerman, R., Tripakis, S.: Modular Code Generation from Triggered and
Timed Block Diagrams. In: IEEE Real-Time and Embedded Technology and Ap-
plications Symposium, pp. 147-158 (2008)
8. Sontag, E.D.: Mathematical Control Theory: Deterministic Finite Dimensional
Systems. Second Edition, Springer, New York (1998)
9. Proakis, J., Manolakis, D.: Digital Signal Processing: Principles, Algorithms and
Applications, 4th ed. Pearson (2006)
10. Tiwari, A.: Formal semantics and analysis methods for Simulink Stateflow models.
Unpublished report, SRI International (2002)
An Abstract Block Formalism for Engineering Systems 463
11. Bouissou, O., Chapoutot, A.: An operational semantics for Simulink’s simulation
engine. LCTES 2012, pp. 129-138. (2012)
12. Agrawal, A., Simon, G., Karsai, G.: Semantic translation of Simulink/Stateflow
models to hybrid automata using graph transformations. Electronic Notes in The-
oretical Computer Science 109, 43-56 (2004)
13. Marian, N., Ma, Y.: Translation of Simulink Models to Component-based Software
Models. In: 8-th Int. Workshop on Research and Education in Mechatronics, 14-15
June 2007, Talin University of Technology, Estonia (2007)
14. Pinto, R., Sangiovanni-Vincentelli, A., Carloni L.P., Passerone, R.: Interchange
formats for hybrid systems: Review and proposal. In: HSCC 05: Hybrid Systems
Computation and Control. Springer-Verlag, pp. 526–541 (2005)
15. Beek, D.A., Reniers, M.A., Schiffelers, R.R., Rooda, J. E.: Foundations of a Com-
positional Interchange Format for Hybrid Systems. In: HSCC’07, pp. 587-600
(2007)
16. Henzinger, T.: The theory of hybrid automata. In: IEEE Symposium on Logic in
Computer Science, pp. 278–292 (1996)
17. Goebel, R., Sanfelice, R., Teel, R.: Hybrid dynamical systems. In: IEEE Control
Systems Magazine 29, 29–93 (2009)
18. Schrammel, P., Jeannet, B.: From hybrid data-flow languages to hybrid automata:
a complete translation. In: HSCC 2012: pp. 167–176 (2012)
19. Camhbel, M.,K., Heemels, A.J., van der Schaft, A.J., Schumacher, J.M.: Solution
concepts for hybrid dynamical systems. In: Proc. IFAC 15th Triennial World
Congress, Barcelona, Spain (2002)
20. Nikitchenko, N.S.: A composition nominative approach to program semantics.
Technical report IT-TR 1998-020, Technical University of Denmark, 103 p. (1998)
21. Windeknecht, T.G.: Mathematical systems theory: Causality. Mathematical sys-
tems theory 1, pp. 279-288 (1967)
22. Mesarovic, M.,D., Takahara, Y.: Abstract systems theory. Springer, Berlin Hei-
delberg New York, 439 p. (1989)
23. Foo, N., Peppas, P.: Realization for Causal Nondeterministic Input-Output Sys-
tems. Studia Logica 67, pp. 419-437 (2001)
24. Lin, Y.: General systems theory: A mathematical approach. Springer, 382 p.
(1999)
25. Matsikoudis, E., Lee, E.: On Fixed Points of Strictly Causal Functions. Tech-
nical report UCB/EECS-2013-27, EECS Department, University of California,
Berkeley (2013).
26. Williems, J.: Paradigms and puzzles in the theory of dynamical systems. In: IEEE
Transactions on Automatic Control 36, pp. 259-294 (1991)
27. Williems, J.: On Interconnections, Control, and Feedback. In: IEEE Transactions
on Automatic Control 42, pp. 326-339 (1997)
28. Williems, J.: The behavioral approach to open and interconnected systems. In:
IEEE Control Systems Magazine, pp. 46-99 (2007)
29. Baldwin, C.Y., Clark, K.B.: Design Rules, Volume 1: The Power of Modularity.
MIT Press (2000)
30. Tripakis, S., Lickly, B., Henzinger, T., Lee, E.: On relational interfaces. Proceed-
ings of EMSOFT’2009, pp. 67-76 (2009)
31. Ivanov, Ie.: A criterion for global-in-time existence of trajectories of non-
deterministic Markovian systems. Communications in Computer and Information
Science 347, pp. 111-130, Springer (2012)
32. Carloni, L.P., Passerone, R., Pinto A.: Languages and Tools for Hybrid Systems
Design. Foundations and Trends in Design Automation 1, 1-204 (2006)
Multilevel Environments in Insertion Modeling System
Dmitriy M. Klionov1
1
Kherson State University, 40 rokiv Zhovtnya st. 27, Kherson, Ukraine
soulslayermaster@gmail.com
Abstract. The goal of this paper is to show that the Insertion Modeling Sys-
tem[1] developed by A.A. Letichevsky of the department 100/105 of the Glush-
kov Institute of Cybernetics, National Academy of Science of Ukraine, Kyiv,
Ukraine, can be used as an instrument for the modeling and analysis of complex
distributed systems, such as a client-server architectures. The Insertion Model-
ing [1] is based on the interactions of environments and agents inserted into that
environments. Agents have different behaviors represented as Behavior Alge-
bras, and can also be the environments themselves, having another agents with
different behaviors inserted into them. The definition for multilevel environ-
ments was first given in a paper [1], and was slightly extended in following pa-
pers.
Keywords. Insertion modeling, multilevel environments, compatibility relation,
client-server architecture
Key terms. Computation, Model, Insertion Modeling
1 Introduction
Insertion modeling is a technology for specification and verification of complex dis-
tributed systems based on the interactions of agents and environments. Agents and
environments are models of some entities of real world or components of complex
systems on different levels of abstraction that interact with one another by means of
insertion functions. Also if the environment is considered as an agent it can also be
inserted to other environments. In order to model complex systems those consist of a
lot of components that have hierarchical structure, the notion of multilevel environ-
ments, with agents that are able to move from one environment to another is required.
The notion of mobility of such mobile agents are based on the approach recently fa-
vored in declarative mobile language design is using mobile calculi that extend or
modify the π-calculus [10] with new features, including mechanisms for encryption
and security. Calculi of this kind include, among others, the Spi Calculus [6], and the
Ambient Calculus [7]. In addition, there is a broader body of work favoring declara-
tive approaches, including work in the field of coordination languages. There has also
Multilevel Environments in Insertion Modeling System 465
been a great expansion of the capabilities and security of agent-based languages such
as OAA [10] and D’Agents[13].
According to the Ambient Calculus [7], devised by Luca Cardelli the main diffi-
culty of mobile computations in Web is not in mobility itself but in handling of ad-
ministrative domains. In the early days of the Internet one could rely on a flat name
space given by IP addresses; knowing the IP address of a computer would very likely
allow one to talk to that computer in some way. This is no longer the case: firewalls
partition the Internet into administrative domains that are isolated from each other
except for rigidly controlled pathways. System administrators enforce policies about
what can move through firewalls and how.
The client–server model is the prevalent approach in computer networking. The
model assigns one of two roles to the computers in a network: a client or a server.
A server is a computer system that selectively shares its resources; a client is a com-
puter or computer program that initiates contact with a server in order to make use of
a resource. Data, CPUs, printers, and data storage devices are some examples of re-
sources. This model can be represented as a set of administrative domains, with de-
fined access rules, or as some architectural design pattern, like three-tier pattern. Both
of these are presented in this paper in terms of the insertion modeling.
2 Insertion Modeling System
Insertion modeling system is an environment for the development of insertion ma-
chines and performing experiments with them. Insertion model of a system represent
this system as a composition of environment and agents inserted into it, using the
insertion function. Contrariwise the whole system as an agent can be inserted into
another environment. In this case we speak about the internal and external environ-
ment of a system. Agents inserted into the internal environment of a system them-
selves can be environments with respect to their internal agents. In this case we speak
about multilevel structure of agent or environment and about high level and low level
environments.
Agent and environments have a set of action and a set of behaviors (processes),
defined in behavior algebra. Two set of actions: a set of environment actions and a set
of agent actions define the type of environment. If an agent is about to be inserted into
the environment at least one of its actions must be allowed by this environment. So
the set of agent actions define the type of environments it can be inserted in, as well
as the environment’s set of allowed agent actions define the type of agents that can be
inserted into this environment. Such a relation between types of agents and environ-
ments is called compatibility relation [2], which defines a directed graph. When an
agent is inserted into some environment, it is able to move to another environment if
it is compatible with this environment. For example the rule(1) shows an agent u that
moves to an external environment E, from environment R, it is currently inserted into.
e u
u move
P( E , R, u, move e)
E[ R[u ]] move
E[u , R[]]
_ up ( r e )
(1)
466 D. M. Klionov
Here e and r – are the names of environments, R[] – describes environment R that
currently have no agents inserted into it. Insertion only occurs if a predicate P is true,
and in general case it may depend only on the types of agents and environments. This
example rule shows “one step” movement of an agent u, and if the new state of agent
u has the same type as u, and types of environments E and R had not changed as
well, rule (1) can be considered as commutative. Also“long range” movements can be
defined recursively, for any set of environments between E and R.
3 Insertion Models of Client-Server Architecture
3.1 Domain Model
This model describes a client-server-architecture as a set of administrative domains
that have certain access rules. Each of these domains is represented by an environ-
ment in IMS. Agents are messages that travel over these domains, trying to access
certain protected area of some administrative domain of the server. As an example we
take our website apsystem.or.ua. It is shown at picture below. The top-most environ-
ment E- represents some network (local-area network or internet), with environments
of apsystem itself , and a set of clients C1, C2, … , Cn inserted into the network. Cli-
ent environments create agents and send them over the network in order to gain ac-
cess to some function of apsystem if they have certain permission, or to a domain of
another client. One of the clients can represent a villain (Hacker), which goal is to
find all possible security risks and ways of an attack to curtain security protocol.
In order to access administrative domain and to authorize on a server the client
has to show that it knows some secret, which is only known to client and server (or
two clients that want to exchange some data), and which is not transferred over the
network. This key is used to encode messages (transferred by the agents), and when
agents tries to move into the environment of administrative domain, this key is used to
decode the message, if it is possible than agent inserts into the environment, and pro-
ceeds further. There are many ways for generating such secret.
Fig. 1. The domain model of client server
Multilevel Environments in Insertion Modeling System 467
This model uses the standard Needham–Schroeder Public Key protocol [21]. Each
client and server has a secret key, which is used to decode messages encoded with
appropriate public key. When an agent gets inside the administrative domain (apsys-
tem for example), it have to get a permissions to act inside it. The message transferred
by this agent, contains the information about the access rights of the client who sent it.
This data is used to move further. When an agent reaches some function
(“download_paper” for example) it has a permission to, it is to be sent back by the
server to the client. Account environments that are inserted into the clients and the
top-most environment of the server store all information required to authorize at ap-
propriate client or server. Tables below show all types of environments and agents.
Types of environments of clients and the top-most environment of the server, are
identical. In general the client differs from the server only by the means of environ-
ments inside it, which require an action authorized_move.
Fig. 2. Compatibility graph for the client-server domain model
Vertexes represent agents and environments, and edges represent a compatibility
relation. Directions mean that for example the authorized agent can be inserted into
the environments of the account, server functions environments, clients and servers
environments.
Interactions with agents:
a u
u send
P( E , C , AC , A, u , send a)
a E[C [ AC []], A[u]]
E[C[ AC[u ]], A[]] send (2)
In equation (2) send(A) Means that client C sends the message u, with an appro-
priate account AC, to the server A[], over the network E, where a – is the name of
server A[]. The definition A[] means that there were no agents inserted into this envi-
ronment.
468 D. M. Klionov
Table 1. Actions of agents and environments in domain model
Agent / Envi-
Attributes Actions
ronments type
mb – message body, actual
information carried by this
agent; send a - makes agent to move to the server envi-
ronment named a,
sender – the name of the one
Simple message
who sent this message;
access d - agent tries to authorize in order to
enc_key – key that is used by
enter the environment named d, that is in the
encryption algorithm;
server environment.
mb – message body, actual auth_move d - “authorized move” to some inter-
information carried by this nal environments of the server named d
agent;
get_data(x) - agent shares the data it carries.
invoke(x) - invokes the main function of the
Authorized mes-
environments of the server functions, x – is the
sage
role – defines the access access level of authorized agent. It receives as an
level of this information answer or the result of execution of function, or
the “access denied” message.
done(x) - required to check if the result it carries
is equivalent to the expected result
Clients and the Secretkey – an integer value
top-most envi- of the client’s secret key, that
is used by the Needham- allow(y) - environment checks the incoming
ronment of the
Schroeder algorithm message from the server, y – is the secret key that
server
is used to decode the information from that
Allowed actions:
Nounce – a place for random message.
send, access,
numbers.
authmove
update(x) - account is able to update its data
server – the name of the
about the secret keys used in the Needham-
server it belongs to
Schroeder algorithm
role – an integer value that check_goal(x) - checks if the result brought by
Accounts represent a role of this ac- the message, is equal to the expected result that is
Allowed actions: count at server x
access, authmove, publickey – the public key of
send, get_ data(y), the server, that is used by the
done(z) Needham-Schroeder algo-
create(r,t) - environment creates agent named r,
rithm
which has type t
secret – that will be obtained
by Needham–Schroeder
algorithm
Environments that
represent server check_permission u - checks if the access level
functions( permission – an integer value of agent u is appropriate for performing action, if
download_paper, indicating what the required it do then it is delta, if not then agent receives a
upload_paper) permissions to access it are. message that it has no rights to perform the
Allowed actions: function of this environment
authmove, invoke
E
Allowed actions:
send a
Multilevel Environments in Insertion Modeling System 469
A allow
( y)
A, u access
ca
u
P( A, CA, u, access ca)
A[u, CA[], D[]] A[CA[u ], D[]]
access ca
; (3)
An agent u tries to gain access to the server A[], A tries to authorize it, using the
secret y, if the authorization succeeds, then u enters appropriate account on the server
that is CA, and ca is its name.
CA create
CA, u get
( r ,t )
_ u
data ( x )
P (CA, u , t , get _ data ( x), create( r , t ))
A[CA[u ], D[]] A[CA[u , r ], D[]]
create ( r ,t ) ca
(4)
An account environment CA creates a new agent named r, which type is t . It car-
ries all data received from u, by the action get_data(x), x – is that data. This rule cre-
ates an agent of type authorized_agent , but it can create an agent of any type that is
compatible with this environment.
r r
authmove d
P ( A, CA, D, r , authmove d )
A[CA[u , r ], D[]] A[CA[u ], D [ r ]]]
authmove d
(5)
The authorized agent u moves to the environment D[], that represent one of the
server functions;
D check
D , r invoke
_ permission u
r
P( D, r , check _ permission r , invoke)
D[r ] D [r ]
invoke
(6)
The agent u invokes the main function of D[], and depending on the result of
check_permission u, the result of this invoke might be different.
AC chech
AC r done
_ goal ( x )
( y)
P ( AC , r , check _ goal ( x), done( y ))
AC[r ] AC[ ] (7)
When an agent comes back to the client that sent it, the client checks the message
it carried, and it matches the required result then it is successful termination. These
rules only work if both the client and the server share a secret, known only to them. In
order to safely generate such secret the Needham–Schroeder public key algorithm is
used. Usually the Needham–Schroeder protocol requires a second server that hosts all
the public keys, but for simplicity we assume that all clients and servers know all the
public keys. If the secret has already been created, than it is taken instead of public
key and secret key for encoding and decoding of messages.
It runs as follows:
1. First we check if the secret exists for an account A, if not we send message to the
server A[] by the rule (2), and set the value of an agent’s attribute mb to N1 that is
a simple random number.
2. Then the server A[] uses the rule(3) to decode message using the secret key of
server A[],.if the secret is not created yet.
470 D. M. Klionov
3. Then the server replies by the rule (2) to client C the value of mb is set to (N1,N2),
N1 – is the random number created by the client C, and N2 – is the new random
number.
4. If the first part of the mb is equal to the random number that was generated before,
than C can take the pair (N1,N2), as a secret for the account A.
5. Then C sends a message to A[], that contains N2. When A will receive it, he will
also take the pair (N1,N2), as a secret for account C.
In order to verify this protocol one of the clients has to take the role of a villain, its
goal is to be authorized as another client from the network, using in this case a men-
in-middle attack. [22]
3.2 Insertion Model of Three-Tier Architecture
Unlike of the previous model this one focuses on the actual behavior of data-packages
represented by agents, inside the server environment, divided basically to three layers
according to the three-tier architecture. The example model of the server hosting two
sites apsystem and unarea, is presented.
Fig. 3. Insertion model of three-tier client-server architecture
Their frontends are located inside the presentation tier.
E [Pr[ aps [], un [], [ App [ Php [ aps _ l []], Py [ un _ l []],
(8)
[ Data [ mysql [ aps _ d [], un _ d []]]]]
(8) is the state of environment in such example. E – the top-most environment, Pr-
the presentation tier, aps – the apsystem frontend, un – the unarea frontend, App – the
application tier, PHP\ PY – all sites developed in php and python accordingly,
aps_l\un_l – the logic of apsystem\unarea, Data – the data tier, aps_d\un_d – the data-
base of apsystem\unarea. The user only works with frontend. This means, that the
Multilevel Environments in Insertion Modeling System 471
incoming agent is compatible only with environments of the presentation layer. An
agent inserted into one of the frontends carries one request.
Table 2. Types of agents and environments
Agents/Environments types Actions
execute(x) - executes the request brought by user,
x-is the request data
User request
User_move d - User agent moves to environment,
named d
execute_script(y) - executes the request brought by
Script request
script, y-is the request data
Allowed actions: ex-
Script_move d - script agent moves to environment
cute(x),User_move d
named d
Execute_query(z) - Executes the request brought by
Data base request
data base, z-is the request data
Allowed actions: execute_script(y),
Data_base_move d - Data base agent moves to
Script_move d
environment named d
Environments of the Presentation tier
Allowed actions: exe- Create (r,t) - Creates agent named r, of the type t
cute(x),User_move d, Script_move d
Environments of the Application tier
Allowed actions: execute_script(x), Create (r,t) - Creates agent named r, of the type t
Script_move d,Data_base_move d
Environments of the Data tier
Allowed actions: execute_query(x),
Script_move d, Data_base_move d
Interaction with environments: In the rule (9) u gets inside Pr using user_move pr,
where pr is the name of the environment Pr, if P can allow this. (a simple one step
insertion). The same way u gets inside the environment aps[], using in(aps). This
shows how a user goes to some web-site (apsystem in our case), in order to download
a page for example. In order to do so he has to load a web-page that has a required
link to the paper he wants.
user_ move pr
u
u (9)
user_ move pr)
P(Pr,u, user_ move pr))
E[u, Pr[APS[], App[ APS_ l[],Data[mysql[ APS_ d[]]]]]]
E[Pr[u, APS[], App[ APS_ l[],Data[mysql[ APS_ d[]]]]]]
The link to the paper is stored in the site’s data base that is inside the Data tier,
and the rules for extracting these data, and displaying them is inside the Application
layer. So, the frontend part (environment of apsystem in our case) creates a new
agent, that is compatible only with this environment, and with according environ-
ments of the Application layer:
472 D. M. Klionov
u
execute( x)
(10)
P( APS, u, execute( x))
execute( x)
E[Pr[ APS []],Q]]
E[Pr[APS[u],Q]]
Here we check if u is able to execute its request x if it succeeds than it is DELTA,
if it is NOT able to, than we have to check if there any environments inside aps, that
are compatible with u, go inside them, and again try the same rule. Q is put for sim-
plicity; it describes all rest of environments that are not involved in the current rule. If
there are no such environments or even after insertion to such environment u is still
unable to solve(x), then we have to create a new agent r that will get necessary data
from application tier.
APS create
AP S
( r ,t )
( APS , t , create ( r , t ))
E [Pr[ aps [ u ], Q ]] E [P r [ ap s [ u , r ]], Q ]]
create ( r , t )
(11)
Note: r has to be created inside that environment, which u is currently inserted in.
r execute
_ r
script( y ) (12)
P( APS _ l , r, execute_ script( y))
E[Pr[APS[u], App[ PHP[ APS _ l[r ]]]]] solve
( y)
E [Pr [ APS[u], App[ PHP[ APS _ l [r ]]]]]
The agent r moves to that environment (APS_l). It should go first to the environ-
ment PHP, which is the top-environment of all sites based on PHP, and then moves to
the environment APS_l. The rule for its movement is similar to the movement of a
user request. The execution of script request is different:
(13)
r r
script _ move aps
P ( PR , App , PHP , APS , APS _ l , r , script _ move aps )
E [Pr[ APS [u ], App [ PHP [ APS _ l[ r ]]]]]
script _ move aps
E [P r [ APS [u , r ], Ap p [ PH P [ APS _ l []]]]]
If the execution succeeds, than r moves back to the environment, which created it.
If not, then the environment APS_l, creates a new agent, which is the query for the
data tier. The rules are similar.
4 Conclusions
The client-server model can be considered as a prevalent approach in computer net-
working, and is one of the best examples of complex distributed systems. Two exam-
ples of insertion models of client-server architecture are presented in this paper: the
domain model – as a set of administrative domains with pre-defined access rules; and
a three-tier architecture - a client–server architecture in which the presentation, the
application processing, and the data management functions are logically separated.
Both these insertion models with multilevel environments and mobile agents can be
extended later for more complicated applications, such as the verification of crypto-
Multilevel Environments in Insertion Modeling System 473
graphic protocols, the problem solving, the constraint propagation, the cognitive ar-
chitectures.
References
1. Letichevsky, A. A.: Insertion Modeling. Control Systems and Computers, 6, 3-14 (2012)
2. Letichevsky, A.: Algebra of Behavior Transformations and its Applications. In:
Kudryavtsev, V. B., Rosenberg, I. G. (eds.) Structural Theory of Automata, Semigroups,
and Universal Algebra. NATO Science Series II. Mathematics, Physics and Chemistry,
vol. 207, pp. 241-272, Springer (2005)
3. Baranov, S., Jervis, C., Kotlyarov, V., Letichevsky, A., Weigert, T.: Leveraging UML to
Deliver Correct Telecom Applications. In: L. Lavagno, G. Martin, and B.Selic (eds.) UML
for Real: Design of Embedded Real-Time Systems. Kluwer Academic Publishers, Amster-
dam (2003)
4. Letichevsky, A.A., Kapitonova, J., Letichevsky, A. Jr., Volkov, V., Baranov, S., Kot-
lyarov, V., Weigert, T.: Basic Protocols, Message Sequence Charts, and the Verification of
Requirements Specifications. Computer Networks, (47), 662–675 (2005)
5. Kapitonova, J., Letichevsky, A., Volkov, V., Weigert, T.: Validation of Embedded Sys-
tems. In: R. Zurawski (ed.) The Embedded Systems Handbook. CRC Press, Miami (2005)
6. Abadi, M., Gordon, A.D.: A Calculus for Cryptographic Protocols: the spi Calculus. In:
Proc. 4th ACM Conference on Computer and Communications Security, pp. 36-47, (1997)
7. Cardelli., L., Gordon, A.: Mobile Ambient. In: Nivat, M. (ed.) Proc. FoSSaCs’98: Founda-
tions of Software Science and Computational Structures, LNCS 1378, pp. 140–155,
Springer-Verlag (1999)
8. Lange, D.B., Oshima: Programming and Deploying Java Mobile Agents with Aglets. Ad-
dison-Wesley (1998)
9. Kotz, D., Gray, R.S.: Mobile Agents and the Future of the Internet. ACM Operating Sys-
tems Review 33(3), 7–13, (1999)
10. Milner, R., Parrow, J., Walker, D.: A Calculus of Mobile Processes (Parts I and II). Infor-
mation and Computation 100, pp. 1-77 (1992)
11. Martin, D., Cheyer, A., Morgan, D.: The Open Agent Architecture: A Framework for
Building Distributed Software Systems. Applied Artificial Intelligence, 12, 91–128 (1999)
12. Gray, R.S., Kotz, D., Cybenko, G., Rus, D.: D’Agents: Security in a Multiple-Language,
Mobile-Agents System. In: Vigna G. (ed.) Mobile Agents and Security, LNCS 1419, pp.
154–187, Springer-Verlag (1998)
13. Milner, R.: A Calculus of Communicating Systems, LNCS 92, Springer-Verlag, (1980)
14. Milner, R.: Communication and Concurrency. Prentice Hall, (1989).
15. Milner, R.: The Polyadic π-Calculus: a Tutorial. Tech. Rep. ECS–LFCS–91–180, Labora-
tory for Foundations of Computer Science, Department of Computer Science, University
of Edinburgh, UK (1991)
16. Park, D.: Concurrency and Automata on Infinite Sequences. In: LNCS 104. Springer-
Verlag, (1981)
17. Roggenbach, M., Majster-Cederbaum, M.: Towards a Unified View of Bisimulation: a
Comparative Ctudy. TCS, 238, 81–130 (2000)
18. ITU-T. Z.120 Recommendation Z.120 (11/99): Languages for telecommunications appli-
cations – Message Sequence Charts (MSC) (1999)
19. ITU-T. Z.100 Recommendation Z.100 – Specification and Description Language (SDL)
(1999)
474 D. M. Klionov
20. Rutten, J.: Coalgebras and Systems. TCS, 249
21. Needham, R., Schroeder, M.: Using Encryption for Authentication in Large Networks of
computers. Comm. ACM, 21(12), 993–999 (1978)
22. Lowe, G.: An Attack on the Needham-Schroeder Public Key Authentication Proto-
col. Information Processing Letters, 56(3), 131–136 (1995)
23. Eckerson, W.: Three Tier Client/Server Architecture: Achieving Scalability, Performance,
and Efficiency in Client Server Applications. Open Information Systems. 3(20), 10, 1
(1995)
Clocks Model for Specification and Analysis
of Timing in Real-Time Embedded Systems
Iryna Zaretska1 , Galyna Zholtkevych1 , Grygoriy Zholtkevych1
and Frédéric Mallet2
1
V.N. Karazin Kharkiv National University,
School of Mathematics and Mechanics, 4, Svobody Sqr., 61022, Kharkiv, Ukraine
zar@univer.kharkov.ua,{galyna,g}.zholtkevych@gmail.com
2
Université Nice Sophia Antipolis, AOSTE Team Project (INRIA/I3S),
INRIA Sophia Antipolis Méditerranée, 2004 rte des Lucioles (Lagrange L-043) BP93,
F-06902 Sophia Antipolis Cedex, France
frederic.mallet@unice.fr
Abstract. Problems concerning formal semantics for Clock Constraint
Specification Language (CCSL) are considered in the paper. CCSL is
intended for describing logical time models for real-time embedded sys-
tems and the language is a part of UML profile for MARTE. There exist
two approaches to introduce a denotational semantics for CCSL. A pure
relational subset of CCSL is defined in the paper. The notion of time
structure with clocks is introduced to refine describing denotational se-
mantics for this CCSL subset, which authors called RCCSL. Semantic
properties of RCCSL have been studied. Theorem about coincidence se-
mantics of RCCSL for the two approaches is proved.
Keywords. Embedded system, real-time system, time modelling, time
structure, clock constraint, formal specification
Key terms. ConcurrentComputation, FormalMethod, SpecificationPro-
cess, VerificationProcess, MathematicalModeling
1 Introduction
Nowadays, the growth of using distributed real-time systems (including embed-
ded systems) [4] is the developing trend for Information and Communication
Technology. There are two reasons for such growth: first, the physical limit for
processor acceleration is reached, and, second, using mobile and cloud technolo-
gies are explosively expanded. The impossibility to continue over-clocking of a
processor leads to using a multi-core system, which is parallel and distributed.
A complex consisting of a computational cloud and an ensemble of mobile de-
vices is a parallel and distributed system too. Moreover its structure is not fixed.
476 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
Each of the cases requires using different kinds of multiprocessing architectural
and software solutions [3]. Therefore, providing correct working of such systems
requires more research in the area.
Mathematical modelling of systems makes possible to develop formal specifi-
cations and methods of their analysis as a base for trustworthy system construct-
ing. There are a lot of approaches to modelling multiprocessor systems. First of
all, the following ones should be noticed: CSP of C.A.R. Hoare [8], π-calculus of
R. Milner [11], abstract state machine model [5], and processing algebra [14].
This paper is devoted to formal methods for an important subclass of mul-
tiprocessing distributed systems, namely, real-time embedded (RTE) systems.
These methods are closely connected with the UML profile for MARTE (Mod-
elling and Analysis of Real-Time and Embedded systems) [2, 15]. In the context
of the MARTE approach UML [16, 17] is used to build engineering models of a
developing system. But the UML notation does not support detailed description
of interactions for joining components into a united RTE system. A very com-
mon way to specify conditions for the system integrity is through the Object
Constraint Language (OCL) [12]. However, no facilities for specifying temporal
constraints are provided by the OCL standard. The Clock Constraint Specifica-
tion Language (CCSL) [2] was defined in an annex of MARTE as a way to build
logical and temporal constraints on model elements.
CCSL is intended to describe the temporal ordering of interactions between
components of a distributed software system. It focuses on the ordering of event
occurrences, but not on their chronometric characteristics. It relies on a logical
time model inspired by the work on synchronous systems and their polychronous
extensions.
The denotational semantics for basic constructions of CCSL is given in [10].
It is based on the notion of a time structure with clocks, other approach [1]
defines an operational way to compute runs for CCSL specifications. The main
contribution of this paper is a demonstration that the relationship of semantics
consequence based on time structures as models of constraints and semantics
consequence based on time structures associated with runs are only equivalent
for a subset of CCSL, which we call RCCSL.
2 Syntax of Pure Relational CCSL
In the paper we restrict ourself to a very simple sublanguage of CCSL, which
we call the pure relational CCSL (RCCSL). Syntax of this subset is given here
using EBNF [6].
clock constraint =
clock relation, {’,’, clock relation};
clock relation =
clock reference, sign of clock relation, clock reference;
sign of clock relation =
’subclocking’ |
Clocks Model for Specification and Analysis of Timing ... 477
’exclusion’ |
’coincidence’ |
’cause’ |
’precedence’;
clock reference =
? any element of clock set ?;
Below we use the next notation for symbols of clock relations (see Table 1).
Table 1. Symbols of clock relations
Relation name Relation symbol
subclocking ⊂
exclusion #
coincidence =
cause 4
precedence ≺
These five binary relations on a clock set C are determined as logical primi-
tives for CCSL in [1].
Defining semantics for RCCSL is one of the paper objectives. Following the
paper [10], we define the denotational meaning for a set of clock constraints as
some class of time structures expanded by a classification for event occurrences.
The next section is devoted to describing such structures.
3 Time Structure with Clocks
Let consider a set of event occurrences, which is below denoted by I. Elements of
the set I are called instants. Some pairs of instants denotes instant pairs, whose
elements are ordered in time: i1 4 i2 is denoted the fact ”an instant i1 causes
an instant i2 ” or equivalently ”an instant i1 cannot occur later than an instant
i2 ”, where i1 , i2 ∈ I. This relation is called ’cause’. It is naturally to suppose
that cause is a pre-order.
As known [7, section 1.3], each pre-order can be decomposed uniquely into
the union of two relations such that the former is a strict order (it is denoted
bellow by ’≺’ and called a precedence) and the latter is an equivalence (it is
denoted bellow by ’≡’ and called a coincidence). These relations are connected
by the next property:
for any instants i1 , i01 , i2 , i02 ∈ I
the validity of i1 ≡ i01 , i2 ≡ i02 , and i1 ≺ i2 implies (1)
truth of i01 ≺ i02 .
478 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
Moreover, if we have a strict order and an equivalence on the same set and these
relations satisfy (1) then their union is a pre-order.
Now, we can introduce the notion of a time structure for formalising our
understanding a set of instants.
Definition 1. Let (I, 4) be a pair of a set and a pre-order on this set respec-
tively. Denote by ≺ the strict order corresponding to the pre-order 4. The pair
(I, 4) is called a time structure if the next property (the property of cause finite-
ness [13]) holds:
the set {i0 ∈ I | i0 ≺ i} is finite for all i ∈ I. (2)
Definition 1 is based on the corresponding definition in [10]. One can compare
them with the definition of a time structure in [13]. Difference consists in a
possibility of modelling an instant coincidence.
Note that Definition 1 specifies the set of instants and some time relations
on it but it does not determine any classification of instants in compliance with
their sources. Therefore, in the following [2] we introduce such a classification
by adding a finite set of instant sources called clocks and by mapping the set of
instants into this clock set.
Definition 2. Let (I, 4) be a time structure, C be a finite set of clocks, and
π : I → C be a map then the quadruple (I, 4, C, π) is called a time structure with
clocks if the next property holds:
for any clock c ∈ C and i1 , i2 ∈ π −1 (c)
the validity of i1 6= i2 implies truth of i1 ≺ i2 ∨ i2 ≺ i1 , (3)
i.e. π −1 (c) is linearly ordered by the restriction of the cause.
If c ∈ C then the set π −1 (c) is usually denoted by Ic . It can be considered as an
event stream generated by the source associated with the clock c.
From Definition 1 and Definition 2 the next fact follows immediately.
Proposition 1. Let (I, 4, C, π) be a time structure with clocks then
1. Ic is well-ordered by the strict order ≺ for all c ∈ C;
2. ordinal type of Ic for any c ∈ C is less or equal to ω, where ω is the first
infinite ordinal.
Proof. Firstly note that property (3) implies linear ordering Ic for an arbitrary
c ∈ C.
Further, suppose that A is some non-empty subset of Ic for an arbitrary c ∈ C,
i is some element of A.
If for all i0 ∈ A the statement i ≺ i0 ∨ i = i0 is true then inf A = i ∈ A.
If there exists i0 ∈ A such that i0 ≺ iTthen the set A(i) = {i0 ∈ A | i0 ≺ i} is
not empty. It is evident that A(i) = A {i0 ∈ Ic | i0 ≺ i}. This equality and the
property of cause finiteness (2) imply finiteness of A(i). So, taking into account
Clocks Model for Specification and Analysis of Timing ... 479
property (3) we can conclude that A(i) is a finite linearly ordered set. Hence,
there exists i∗ ∈ A(i) such that i∗ = inf A(i). It is evident that
inf A = inf A(i) = i∗ ∈ A(i) ⊂ A.
Thus, inf A ∈ A and Ic is well-ordered.
The supposition that ordinal type of Ic for some c ∈ C is greater than ω is
inconsistent with the property of cause finiteness (2). t
u
Corollary 1. Any instant i ∈ I is uniquely determined by the pair (π(i), idx(i)),
which is an element of the set C × N. Here, idx is a map from I into N such that
idx(i) = |{i0 ∈ Iπ(i) | i0 ≺ i}| + 1,
where the number of elements in a set A is denoted by |A|.
The designation TC is used below to refer to the class of time structures with
C as a set of clocks.
Remark 1. One can show that this class is a set but we do not do it in the paper.
4 Denotational Semantics for RCCSL
Usually, a denotational semantics can be considered as the theory of models for
the corresponding language. We shall use time structures with clocks as models
for describing meaning of clock constraints.
4.1 Some General Notes
One can identify a class of event occurrences of the same type with a set of
instants for some clock in the process of specifying interactions between compo-
nents of distributed parallel systems. Such an identification is provided by fixing
a set of clocks C and describing rules of interacting system components. These
rules divide the set TC into two subsets: the subset of time structures satisfying
the constraints and the set of time structures contradicting them. Taking into
account the specification of RCCSL one can say that a clock constraint is a finite
set of clock relations. If the set of clock relations determining the constraint is
denoted by C then the fact ”the time structure T ∈ TC satisfies the constraint
C” can be written as T |= C. More precisely, T |= C means that for each C ∈ C
the clause T |= C is true.
Further, for a constraint C, JCK denote the following set {T ∈ TC | T |= C}.
The first important problem is the consistency problem for the constraint.
The rigorous problem formulation has usually the form:
Problem 1 (Consistency Problem). For a constraint C check that the set JCK is
not empty.
480 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
The second important problem is the semantic consequence for the constraints.
The rigorous problem formulation has the next form:
Problem 2 (Semantic Consequence Problem). For a constraint C and a clock
relation r check that JCK ⊂ JrK (or in the another notation C r).
Below we use the notation {C} for the set of clock relations that form the
constraint C. It is easy to see that the next properties of the relationship are
true.
Proposition 2. The next properties are satisfied:
1. if a constraint C and a clock relation r satisfy the condition r ∈ {C} then
C r;
2. if constraints C1 and C2 and a clock constraint r satisfy the next condition
C1 r0 for all r0 ∈ {C2 } and C2 r are true then C1 r is true.
Proof is omitted t
u
To complete defining the denotational semantics for RCCSL we should determine
the meaning of basic clock relations.
4.2 Subclocking
This relation is intended for specifying a requirement to synchronize each instant
of one clock with some instant of an other clock. In this case the first clock is
called a subclock of the second clock.
More precisely, let c0 , c00 ∈ C and T ∈ TC then T |= c0 ⊂ c00 means that
there exists a strict monotonic map h : Ic0 → Ic00 such that i ≡ h(i) for any
i ∈ Ic0 .
Proposition 3 (Trivial Subclocking). For each c ∈ C the clause c ⊂ c is
true.
Proof is trivial t
u
Proposition 4 (Transitivity Law for Subclocking). For each c0 , c00 , c000 ∈ C
the clause c0 ⊂ c00 , c00 ⊂ c000 c0 ⊂ c000 is true.
Proof. Let hc00 c0 : Ic0 → Ic00 , hc000 c00 : Ic00 → Ic000 be strict monotonic maps pro-
viding the validity of the clauses T |= c0 ⊂ c00 and T |= c00 ⊂ c000 respectively
for some T . It is easy to see that the map hc000 c00 ◦ hc00 c0 provides the validity of
the clause T |= c0 ⊂ c000 t
u
4.3 Exclusion
This relation is used for specifying the mutual exclusion for two events.
More formally, let c0 , c00 ∈ C and T ∈ TC then T |= c0 # c00 means that for
any i0 ∈ Ic0 , i00 ∈ Ic00 the coincidence i0 ≡ i00 is false.
Clocks Model for Specification and Analysis of Timing ... 481
Proposition 5 (Irreflexivity Law for Exclusion). For each c ∈ C the equal-
ity Jc # cK = ∅ is true.
Proof is trivial t
u
Proposition 6 (Symmetry Law for Exclusion). For each c0 , c00 ∈ C the
clause c0 # c00 c00 # c0 is true.
Proof is trivial t
u
4.4 Coincidence
This relation describes synchronization of two event sources.
More precisely, let c0 , c00 ∈ C and T ∈ TC then T |= c0 = c00 means that
there exists a strict monotonic bijection h : Ic0 → Ic00 such that i ≡ h(i) for any
i ∈ Ic0 .
Proposition 7 (Trivial Coincidence). For each c ∈ C the clause c = c is
true.
Proof is trivial t
u
Proposition 8 (Symmetry Law for Coincidence). For each c0 , c00 ∈ C the
clause c0 = c00 c00 = c0 is true.
Proof. Let h : Ic0 → Ic00 be a strict monotonic bijection providing the validity
of the clause T |= c0 = c00 for some T and h−1 be its inverse map.
Suppose that i0 , i00 ∈ Ic00 , i0 ≺ i00 , and h−1 (i0 ) 6≺ h−1 (i00 ) then either h−1 (i0 ) =
h−1 (i00 ) or h−1 (i00 ) ≺ h−1 (i0 ). But the first alternative contradicts to bijectivity
of h, and the second alternative and strict monotonicity of h implies i00 ≺ i0 . The
last clause contradicts to irreflexivity of the precedence relation. These contra-
dictions show that h−1 is a strict monotonic map.
Further, for any i ∈ Ic00 we have that h−1 (i) ∈ Ic0 and h−1 (i) ≡ h(h−1 (i)) = i.
Thus, the clause T |= c00 = c0 is true t
u
Proposition 9 (Transitivity Law for Coincidence). For each c0 , c00 , c000 ∈ C
the clause c0 = c00 , c00 = c000 c0 = c000 is true.
Proof is similar to proof of Proposition 4 t
u
4.5 Cause
This relation is intended for specifying that each instant of one clock is caused
by an instant in another clock.
More precisely, let c0 , c00 ∈ C and T ∈ TC then T |= c0 4 c00 means that
there exists a strict monotonic map h : Ic00 → Ic0 such that h(i) 4 i for any
i ∈ Ic00 .
Proposition 10 (Trivial Cause). For each c ∈ C the clause c 4 c is true.
482 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
Proof is trivial t
u
Proposition 11 (Transitivity Law for Cause). For each c0 , c00 , c000 ∈ C the
clause c0 4 c00 , c00 4 c000 c0 4 c000 is true.
Proof is similar to proof of Proposition 4 t
u
4.6 Precedence
This relation is a stronger variant of the cause relation.
Namely, let c0 , c00 ∈ C and T ∈ TC then T |= c0 ≺ c00 means that there exists
a strict monotonic map h : Ic00 → Ic0 such that h(i) ≺ i for any i ∈ Ic00 .
Proposition 12 (Irreflexivity Law for Precedence). For each c ∈ C the
equality Jc ≺ cK = ∅ is true.
Proof is trivial t
u
Proposition 13 (Transitivity Law for Precedence). For each
c0 , c00 , c000 ∈ C the clause c0 ≺ c00 , c00 ≺ c000 c0 ≺ c000 is true.
Proof is similar to proof of Proposition 4 t
u
4.7 Interdependencies Laws for the Basic Relations
Above we considered properties of each basic relation but interdependencies
between these relations were not in our focus. Thus, such interdependencies are
considered below. The next lemma is needed to ground these dependencies.
Lemma 1. Let (X, ≤) be a well-ordered set and φ : X → X be a strict mono-
tonic map such that for all x ∈ X the assertion φ(x) ≤ x is true then φ is the
identity map.
Proof. One can prove the lemma by using the transfinite induction t
u
Proposition 14 (Interdependencies Laws for the Basic Relations).
1. For each c0 , c00 ∈ C the clause c0 ⊂ c00 , c00 ⊂ c0 c0 = c00 is true.
2. For each c0 , c00 ∈ C the clock relations c0 ⊂ c00 and c0 # c00 are inconsistent,
i.e. Jc0 ⊂ c00 , c0 # c00 K = ∅.
3. For each c0 , c00 ∈ C the clause c0 ⊂ c00 c00 4 c0 is true.
4. For each c0 , c00 ∈ C the clause c0 4 c00 , c00 4 c0 c0 = c00 is true.
Proof. 1) For any T ∈ TC the validity of the assertion ”T |= c0 = c00 implies
T |= c0 ⊂ c00 ” is evident.
Let’s check the validity of the inverse assertion. Denote the strict monotonic maps
that provide for some T ∈ TC the validity of T |= c0 ⊂ c00 and T |= c00 ⊂ c0
Clocks Model for Specification and Analysis of Timing ... 483
by hc00 c0 : Ic0 → Ic00 and hc0 c00 : Ic00 → Ic0 respectively. We claim that they are
mutually inverse.
Indeed, for any i ∈ Ic0 we have the next coincidences: i ≡ hc00 c0 (i) and hc00 c0 (i) ≡
hc0 c00 (hc00 c0 (i)). These coincidences and the Transitivity Law for Coincidence (see
Proposition 9) provide the validity of the coincidence i ≡ hc0 c00 (hc00 c0 (i)). Taking
into account that both i and hc0 c00 (hc00 c0 (i)) are elements of Ic0 and the fact that
restriction of 4 on Ic0 is a strict order (see Proposition 1) one can derive the
equality i = hc0 c00 (hc00 c0 (i)).
The equality i = hc00 c0 (hc0 c00 (i)) for all i ∈ Ic00 is derived similarly. Thus, hc00 c0
is a bijection.
2) Proof is trivial.
3) Proof is trivial.
4) Really, let hc00 ,c0 : Ic0 → Ic00 and hc0 ,c00 : Ic00 → Ic0 be strict monotonic
maps provided for some T ∈ TT the validity of the clauses T |= c00 4 c0 and
T |= c0 4 c00 respectively. Then the map φ = hc0 ,c00 ◦ hc00 ,c0 : Ic0 → Ic0 is
strict monotonic and it satisfies the condition φ(i) 4 i. Therefore, applying the
Lemma 1 allows to conclude that φ and the identity map are equal t
u
5 Runs and Chronometers
Following [1], in this section we introduce the notion of a run for a set of clocks.
We use this notion to define a behavioural model for the set of clocks.
Definition 3 (see [1]). Let C be a finite set of clock then any map r : N → 2C
such that r(t) = ∅ implies r(t0 ) = ∅ for all t0 > t is called a run for C.
This definition means that if r is a run then at the (global) time t all clocks of
the set r(t) and only them are triggered.
For each run r one can construct a quadruple T [r] = (Ir , 4, C, πr ) by the
following way:
– Ir = {(c, t) ∈ C × N | c ∈ r(t)};
– (c0 , t0 ) 4 (c00 , t00 ) if and only if t0 ≤ t00 ;
– πr (c, t) = c for all (c, t) ∈ Ir .
Proposition 15. T [r] is a time structure with clocks for given run r.
Proof. It is proved by trivial checking properties (2) and (3) t
u
Hence, we can define the semantic relationship between a run r and a constraint
C by the next way: r |= C if and only if the clause T [r] |= C is true. Also, we can
introduce the relationship C1 run C2 as an abbreviation of the sentence ”for
any r such that r |= C1 the next relationship r |= C2 is valid”.
Proposition 15 allows to suggest that a run carries more information than a time
structure because a run depends on global time. A refinement and a substanti-
ation of this hypothesis is discussed below.
The notion of chronometer is introduced to specify dependences between time
structures and runs.
484 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
Definition 4. Let T = (C, I, 4, π) be a time structure with clocks and χ : I → N
be a map such that the next assertions are true:
for any i0 , i00 ∈ I the coincidence i0 ≡ i00 implies χ(i0 ) = χ(i00 ) (4)
for any i0 , i00 ∈ I the strict precedence i0 ≺ i00 implies χ(i0 ) < χ(i00 ) (5)
0 0
for any t, t ∈ N the validity of the clauses t ∈ χ(I) and t < t
(6)
implies truth of the clause t0 ∈ χ(I)
then χ is called a chronometer on T [9].
Example 1. Let C be a finite set of clocks, r be a run for C. Then it is evident that
the map χ∗ : Ir → N determined by the equality χ∗ (c, t) = t is a chronometer.
Hence, Example 1 shows that each time structure generated by a run has a
native chronometer χ∗ .
Proposition 16. Let T be a time structure with clocks and χ : I → N be a
chronometer then the map r[T , χ] : N → 2C defined by the next formula
r[T , χ](t) = π(χ−1 (t)) (7)
is a run.
Proof. To prove the proposition we should show that r[T , χ](t) = ∅ for some
t ∈ N implies r[T , χ](t0 ) = ∅ for any t0 ≥ t.
Suppose existence of t1 and t2 such that t1 < t2 , π(χ−1 (t1 )) = ∅, but π(χ−1 (t2 )) 6=
∅. Taking into account this assumption one can derive that χ−1 (t1 ) = ∅ and
χ−1 (t2 ) 6= ∅. Hence, t1 ∈/ χ(I) and t2 ∈ χ(I). We have obtained the contradic-
tion to condition (6) of Definition 4 t
u
The next property for the chronometer χ∗ from Example 1 holds.
Proposition 17. Let r be a run for a clock set C then the next equality holds
r[T [r], χ∗ ] = r. (8)
Let T = (C, I, 4, π) be a time structure with clocks and χ : I → N be a chronome-
ter on T then the map χ b : I → C × N defined in the next way χ
b(i) = (π(i), χ(i))
is a map onto Ir[T ,χ] such that any coincidence i0 ≡ i00 implies the coinci-
dence χb(i0 ) ≡ χ
b(i00 ) in T [r] and any precedence i0 ≺ i00 implies the precedence
0 00
b(i ) ≺ χ
χ b(i ) in T [r].
Proof. Really,
r[T [r], χ∗ ](t) = πr (χ−1
∗ (t)) =
πr ({(c, t) ∈ Ir }) = πr ({(c, t) ∈ C × N | c ∈ r(t)}) = r(t).
Further, (c, t) ∈ Ir[T ,χ] if and only if c ∈ r[T , χ](t). It is easy to see that the last
clause is equivalent to existence of i ∈ I such that c = π(i) and t = χ(i), i.e. it
is equivalent to (c, t) = χ b(i).
If i0 ≡ i00 then χ(i0 ) = χ(i00 ) by definition of a chronometer, hence χ b(i0 ) ≡ χ
b(i00 ).
0 00 0 00 0 00
Similarly, if i ≺ i then χ(i ) < χ(i ), therefore χ b(i ) ≺ χ b(i ) t
u
Clocks Model for Specification and Analysis of Timing ... 485
Proposition 18. There exists only one chronometer on T [r] for any run r.
Proof. For any run r there exists the chronometer χ∗ on T [r]. Let χ be an
other chronometer on T [r]. For (c0 , t), (c00 , t) ∈ I[r] using (4) we have χ(c0 , t) =
χ(c00 , t). Hence, taking into account Definition 3 one can obtain that χ(c, t) = τ (t)
where τ is strict monotonic function from α into α for some cardinal α ≤ ω. Thus,
τ is the identity function and χ = χ∗ t
u
Hence, a chronometer exists on a time structure associated with a run. We claim
that a chronometer exists on any time structure with clocks.
The next binary relation / on a time structure with clocks will be used for
describing an algorithm that calculates timestamps for instants. More precisely,
if i0 , i00 ∈ I then i0 / i00 means that for all i ∈ I the validity of the next clause
i ≺ i00 & i0 4 i implies truth of the coincidence i ≡ i0 . It is easy seen that if
i1 ≡ i01 , i2 ≡ i02 , and i1 / i2 then i01 / i02 .
Now we can construct the algorithm that allows to calculate timestamps
for instants on an arbitrary time structure with clocks. This Algorithm 1 is a
generalization of Lamport’s algorithm [9].
Algorithm 1: Computing timestamp for an instant
input : T = (C, I, 4, π) is a time structure with clocks,
i is an element of I
output: timestamp for the instant i
1 begin
2 count ← 1; D ← ∅; W ← ∅ ; // -- initializing work variables --
3 while i ∈
/ D do // -- main loop --------------------
4 W+ ← {j ∈SI | j ∈
/ D & idx(j) = count};
5 W+ ← WS + / D & (∃j 0 ∈ W+ )j 0 ≡ j};
{j ∈ I | j ∈
6 W ← W W+ ;
0 0 0
7 D+ ← {j S∈ W | (∀j ∈ I)(j / j ⇒ j ∈ D)};
8 D ← D D+ ;
9 W ← W \ D+ ;
10 count ← count + 1;
11 end
12 return count;
13 end
Theorem 1 (existence of a chronometer). Let T be a time structure with
clocks and χ0 : I → N be the function calculated by Algorithm 1 then χ0 is a
chronometer on T .
Proof. One can see that Algorithm 1 builds two sequences of sets
D0 ⊂ D1 ⊂ D2 ⊂ · · · ⊂ Dn ⊂ . . .
W0 , W1 , W2 , . . . , Wn , . . .
486 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
in accordance to the following computational scheme:
W0 = ∅
D0 =∅
0 0 0
S
W n+1 = (WnS {j ∈ I | (∃j ∈ I)(j ≡ j & idx(j ) = n + 1)}) \ Dn
0 0 0
Dn+1 = Dn {j ∈ Wn+1 | (∀j ∈ I)(j / j ⇒ j ∈ Dn )}
and maps an instant i ∈ I into χ0 (i) = inf{n ∈ N | i ∈ Dn }.
Firstly, note that supposition about partial definiteness of χ0 implies existence of
an infinite sequence i1 . i2 . . . . . But it contradicts the causes finiteness property
(2).
Secondly, it is true by the construction of Dn that the validity of i0 ≡ i00 implies
the truth of the following statement: i0 ∈ Dn if and only if i00 ∈ Dn . Hence, we
obtain that i0 ≡ i00 implies χ0 (i0 ) = χ0 (i00 ).
Further, similar reasoning provides the validity of the following statement: i0 ≺ i00
implies χ0 (i0 ) < χ0 (i00 ).
Finally, the simple inequality idx(i) ≤ χ(i), which is correct for any i ∈ I and
any chronometer χ on T , provides the validity of property (6) t
u
Corollary 2. There exists a chronometer on an arbitrary time structure with
clocks.
6 Equivalence of Semantics for RCCSL Determined by
Relations and run
In the section the notion of a chronometer is used to prove the theorem about
equivalence of the relationships and run . The theorem is the main result of
the paper. Taking into account the theorem one can confine himself to checking
semantic consequence by using runs. This opens a way to constructing an opera-
tional semantics of RCCSL so that it is equivalent to the denotational semantics
defined above.
We need two lemmas to prove the main theorem.
Let’s use the notation i1 k i2 for instants i1 and i2 such that i1 64 i2 & i2 64 i1 .
Lemma 2. Let T = (C, I, 4, π) by a time structure with clocks and i1 , i2 be
instants such that the clause i1 k i2 is true then there exists a chronometer χ on
T satisfied the following condition χ(i1 ) < χ(i2 ).
Proof. Let’s consider the quadruple T 0 = (C, I, 40 , π) such that i0 ≺0 i00 is valid
if one of the next conditions is true
1. i0 = i1 and i00 = i2 ;
2. i0 ≺ i00 ;
3. i0 ≺ i1 and i2 ≺ i00 ;
and i0 40 i00 if and only if i0 ≡ i00 or i0 ≺0 i00 . It is easy seen that the relation
40 is a pre-order. More over, it satisfies properties (2) and (3). Hence, T 0 is
Clocks Model for Specification and Analysis of Timing ... 487
a time structure with clocks. Using Corollary 2 we obtain that there exists a
chronometer χ on T 0 . But then χ is a chronometer on T and χ(i1 ) < χ(i2 ) is
true t
u
Corollary 3. Let T = (C, I, 4, π) be a time structure with clocks, i0 , i00 ∈ I be
instants, then
1. i0 ≺ i00 is valid if and only if for any chronometer χ on T the inequality
χ(i0 ) < χ(i00 ) is true;
2. i0 ≡ i00 is valid if and only if for any chronometer χ on T the equality
χ(i0 ) = χ(i00 ) is true.
Lemma 3. Let T = (C, I, 4, π) be a time structure with clocks, ∗ be an arbi-
trary sign of a clock relation, c0 and c00 be clocks then T |= c0 ∗ c00 if and only
if r[T , χ] |= c0 ∗ c00 for any chronometer χ on T .
Proof. It is evident that T |= c0 ∗ c00 implies r[T , χ] |= c0 ∗ c00 for any
chronometer χ on T . Hence, we need to prove the inverse statement.
1) Suppose that r[T , χ] |= c0 ⊂ c00 for any chronometer χ on T . Then for any
i ∈ Ic0 and for each chronometer χ there exists an instant iχ ∈ Ic00 such that
χ(i) = χ(iχ ). Denote by X the set formed all iχ . It is a nonempty subset of
Ic00 . Suppose that there exists at least two different elements in the set X. Let’s
denote them by iχ1 and iχ2 . Taking in account linearity of the order on Ic0 and
iχ1 6= iχ2 one can suppose that iχ1 ≺ iχ2 . Therefore χ1 (i) = χ1 (iχ1 ) < χ1 (iχ2 ).
Thus, one of the two cases is realised: i ≺ iχ2 or i k iχ2 . But in the first case
we obtain the inequality χ2 (i) < χ2 (iχ2 ), which contradicts to the choice of iχ2 .
Hence, i k iχ2 is true. Similarly, one can obtain that i k iχ1 is true. Therefore,
we proved that |X| > 1 implies i k iχ for all i ∈ Ic0 and any chronometer χ.
Let i∗ = inf iχ then i k i∗ and χ(i∗ ) ≤ χ(iχ ) = χ(i). This is a contradic-
χ∈X
tion because Lemma 2 provides existence of some chronometer χ0 such that
χ0 (i∗ ) > χ0 (i). Hence, X contains only one element, which we denote by h(i).
By construction we have χ(i) = χ(h(i)) for any chronometer χ. The last prop-
erty implies strict monotonicity of h and the coincidence i ≡ h(i). Therefore,
T |= c0 ⊂ c00 .
2 and 3) Suppose that r[T , χ] |= c0 ∗ c00 for any chronometer χ on T then it is
evident that T |= c0 ∗ c00 where ∗ equals to # or = .
4 and 5) Suppose that r[T , χ] |= c0 ∗ c00 for any chronometer χ on T where ∗
equals to 4 or ≺ . Similarly, in the first case one can derive that T |= c0 ∗ c00
is true t
u
Theorem 2 (about equivalence of semantics). Let C be an arbitrary finite
set of clocks, C1 and C2 be RCCSL constraints then the C1 C2 is true if and
only if C1 run C2 is true.
Proof. One can easily see that the Theorem is the direct consequence of the
Lemma 3 t
u
488 I. Zaretska, G. Zholtkevych, Gr. Zholtkevych and F. Mallet
7 Conclusion
In the paper we have considered the pure relational subset of CCSL (RCCSL)
and have introduced semantics for it by using a class of mathematical objects
called by authors time structures with clocks.
We have studied semantic properties of RCCSL (see Propositions 3 – 9). We hope
that these properties can be a background of an axiomatic basis for analysing
relational clock constraints.
Further we have introduced the notions ”a run” and ”a chronometer”. It allowed
us to study interrelations between time structures and runs, to introduce the
alternative semantics closer to the operational approach than the denotational
semantics discussed earlier.
Finally, the main theorem about equivalence of these two semantics (see Theo-
rem 2) has been proved.
We are planning to continue our research in the next areas:
– building an axiomatic theory of the semantic consequence for RCCSL con-
straints;
– extending results on complete CCSL;
– studying an operational semantics of CCSL and specifying its interrelations
to the denotational semantics.
References
1. André, C.: Syntax and Semantics of the Clock Constraint Specification Lan-
guage (CCSL). Technical report, RR-6925, INRIA (2009), http://hal.inria.fr/inria-
00384077/en/
2. André, C., Mallet, F., de Simone, R.: The Time Model of Logical Clocks available in
the OMG MARTE profile. In: Shukla, S.K., Talpin, J.-P. (eds.) ”Synthesis of Em-
bedded Software: Frameworks and Methodologies Correctness by Construction”,
pp. 201–227. Springer Science+Business Media, LLC New York (2010)
3. Baer, J.-L.: Multiprocessing Systems. IEEE Trans. on Computers. 12, vol. C-25,
1271–1277 (1976)
4. Bonomi, F., Milito, R., Zhu, J., Addepalli, S.: Fog computing and its role in the
internet of things. In: Proceedings of the first edition of the MCC workshop on
Mobile cloud computing, pp. 13–16. ACM New York, NY, USA (2012)
5. Börger, E., Stärk, R.: Abstract State Machines: A Method for High-Level System
Design and Analysis. Springer-Verlag, Berlin Heidelberg (2003)
6. Information technology – Syntactic metalanguage – Extended BNF. ISO/IEC
14977:1996(E)
7. Harzheim, E.: Ordered Sets. Springer Science+Business Media, Inc. New York
(2005)
8. Hoare, C.A.R.: Communicating Sequential Processes. Prentice Hall International
(1985)
9. Lamport, L.: Time, Clocks, and the Ordering of Events in a Distributed System.
Comm. ACM. 7, vol. 12, 558–565 (1978)
10. Mallet, F.: Logical Time @ Work for the Modeling and Analysis of Embedded
Systems, Habilitation thesis. LAMBERT Academic Publishing (2011)
Clocks Model for Specification and Analysis of Timing ... 489
11. Milner, R.: Communicating and Mobile Systems: The Pi Calculus. Cambridge Uni-
versity Press, Cambridge (1999)
12. Information technology – Object Management Group – Object Constraint Lan-
guage (OCL). ISO/IEC 19507:2012(E)
13. Nielsen, M., Plotkin, G., Winskel, G.: Petri nets, event structures and domains.
Theor. Comp. Sc. 1, vol. 13, 85–108 (1981)
14. Process Algebra for Parallel and Distributed Processing. Alexander, M., Gardner,
W. (eds), CRC Press (2009)
15. UML Profile for MARTE: Modeling and Analysis of Real-Time Embedded Sys-
tems. OMG (2011), http://www.omg.org/spec/MARTE/1.1/pdf/
16. OMG Unified Modeling LanguageTM (OMG UML), Infrastructure. OMG (2011),
http://www.omg.org/spec/UML/2.4.1/Infrastructure
17. OMG Unified Modeling LanguageTM (OMG UML), Superstructure. OMG (2011),
http://www.omg.org/spec/UML/2.4.1/Superstructure
Specializations and Symbolic Modeling
Vladimir Peschanenko1, Anton Guba2 and Constantin Shushpanov3
1
Kherson State University, 27, 40 rokiv Zhovtnya str., Kherson, 73000 Ukraine,
vladimirius@gmail.com
2
Glushkov Institute of Cybernetics of NAS of Ukraine, 40, Glushkova ave., Kyiv, 03680,
antonguba@ukr.net
3
LLC «Information Software Systems», 15, Bozhenko str., Kyiv, 03680 Ukraine,
costa@iss.org.ua
Abstract. We present the technique that allows splitting first-order logic formu-
lae into parts which helps to use the special algorithms of satisfiability checking
and predicate transformer, which are the specializations. We describe the
mathematical description of the algorithm of the constructing specializations
and a few particular approaches to them, which speed up modeling of industrial
models. We prove the correctness of satisfiability and predicate transformer
functions. We consider forward and backward applicability of basic protocols
during symbolic modeling and verification We introduce the examples for each
specialization. We provide the experiments with typical real examples.
Keywords. Symbolic modeling, satisfiability, predicate transformer
Key terms. FormalMethod, MathematicalModeling, SoftwareComponent,
VerificationProcess
1 Introduction
The technique of symbolic verification of requirement specifications of software sys-
tems has shown good results in automatic detection of reachability of deadlocks and
violation of user-defined properties [1]. In previous works [2-4] symbolic models of
systems being transition systems with symbolic states represented by formulae of first
order logic were considered. A relation of transitions between the formulae is deter-
mined and marked by basic protocols, which are considered as actions, performed in
the system. A basic protocol is a formula of dynamic logic
x ( ( x, a ) P ( x, a ) ( x, a )) and it describes local properties of the system in
terms of pre- and postconditions α and β. Both are formulae of first order multisorted
logic interpreted on a data domain, P is a process, represented by means of MSC dia-
Specializations and Symbolic Modeling 491
gram and describes the reaction of a system triggered by the precondition, x is a set of
typed data variables, and a is a set of environment attributes. The general theory of
basic protocols is presented in [5].
A transition is considered as an operator in the space of postcondition formulae.
As the operator transforms one formula to another, in [6] a term “predicate trans-
former” is used. Thus, to compute transitions between the states of such models basic
protocols are interpreted as predicate transformers: for given symbolic state of the
system and given basic protocol the direct predicate transformer generates the next
symbolic state as its strongest postcondition, and the backward predicate transformer
generates the previous symbolic state as its weakest precondition. These concepts
have been implemented in VRS (Verification of Requirement Specifications) system
[7] and IMS (Insertion Modeling System) system [8].
An amount of papers with novel and very efficient techniques for computing satis-
fiability using SAT/SMT has been published in the last years, and some very efficient
SMT tools are now available (e.g., BarceLogic [9], CVCLite/CVC/CVC4 [10,11,12],
DLSAT [13], haRVey [14], MathSAT [15], SDSAT [16], TSAT++ [17], UCLID
[18], Yices [19], Verifun [20], Zapato [21], Z3 [22]). An amount of benchmarks,
mostly derived from verification problems, is available at the SMT-LIB [23]. Work-
shops devoted to SMT and official competitions on SMT tools are run yearly.
All these tools could be configured with the help of many parameters, which
means the usage of some techniques, tactics, heuristics or not, in order to gain in per-
formance. In the paper [24] the algorithm configuration problem is stated as follows:
given an algorithm, a set of parameters for the algorithm, and a set of input data,
found parameter values under which the algorithm achieves the best possible per-
formance on the input data. It gives a possibility of automated tuning of algorithm for
obtaining performance on formulae of some theory.
Usually during modeling of real projects we deal with complex environment states
and simple formulae of basic protocols (pre- and postconditions). It means that we
should check the satisfiability of the conjunction of the environment state and the
precondition formula and transform this whole big formula with the help of predicate
transformer [6]. Obviously, the manipulation with whole formulae is not required for
most of cases.
For example, let i, j : int, f : int int be attributes and f (i ) 0 f (0) 5
j 0 be an environment state, and 1 j : j 1 be a basic protocol. Let’s apply
this basic protocol to the environment state. First, the satisfiability of conjunction of
basic protocol precondition and environment state should be checked:
f (i ) 0 f (0) 5 j 0 . This checking should use the notion of functional sym-
bols: (i 0) ( f (i ) f (0)) . After that we should apply basic protocol postcondi-
tion to conjunction of environment state and precondition (see section Application of
Basic Protocol):
(v : int)( f (i ) 0 f (0) 5 v 0 ( j v 1))
f (i ) 0 f (0) 5 (v : int)(v 0 ( j v 1))
f (i ) 0 f (0) 5 j 1
492 V. Peschanenko, A. Guba and C. Shushpanov
It is known that basic protocol changes attribute j only (see section about predicate
transformers). It means that we could apply basic protocol to small part of environ-
ment state that depends on j, but not to whole environment state formula. In this ex-
ample it could be j 0 only. If there are no predicates in projects ,which could com-
pare values of attribute j with values of other attributes, then we could use some spe-
cial theories for manipulating with such formulae. In this example numeral intervals
could be used for representation of values of attribute j. We call such special theories
Specialization of sat, pt functions according to our general algorithm.
So, the main goal of this paper is to present a mathematical description of algo-
rithm of constructing specializations and a few particular approaches to specialization
which speed up modeling of industrial models. This paper is a continuation of the
[25], where only concrete values as a kind of specialization were described.
In the Section 2 we describe the process of forward application of a basic protocol
with the help of the satisfiability and the forward predicate transformer. In the Section
3 we present an applicability of basic protocols using satisfiability and backward
predicate transformer. The specializations by memory usage and functional symbols
are proposed in the Section 4. The results of experiments are discussed in the Section
5. In the Section 6 we summarize advantages of usage of the specializations and what
could be done in the nearest future.
2 Forward Application of Basic Protocol
Let S(a) be an environment state, x( ( x, a ) P ( x, a ) ( x, a )) be a basic proto-
col, where x – parameters of basic protocol, a – attributes of model,
D ( x, a ) E ( a ) ( x, a ) – conjunction of environment state and precondition of
basic protocol.
At the first step of application of basic protocol satisfiability of conjunction of en-
vironment state and precondition of basic protocol is checked: sat ( D ( x, a)) . If the
formula is unsatisfiable, then basic protocol is not applicable to environment state
S(a). If not, then process P ( x, a) is run and after forward predicate transformer is
applied: pt ( D ( x, a ), ( x, a )) . The process of P ( x, a) is not considered in the paper,
because the specialization tries to speed up the functions sat and pt.
2.1 Satisfiability
The checking formula satisfiability function sat is based on the Shostak method,
adapted to combination of numerical, symbolic and enumerated data types. If all of
the attribute expressions (simple attributes and functional symbols with parameters)
that are free in the formula S are simple, then for satisfiability checking it is sufficient
to prove validity of the closed formula (a, x) D ( x, a ) , where a is a set of all simple
attributes which occur in S, x is a set of parameters of basic protocol. For attribute
expressions with parameters (including access functions to the elements of arrays),
Specializations and Symbolic Modeling 493
the Ackermann reduction of the uninterpreted functional symbols is used, where at-
tribute expression is an attribute or functional symbol with parameters.
The Shostak method consists of the following. An expression of the form f (x) is
called as Functional Expression, if f is an attribute and x is a list of its parameters. At
first, superpositions of functional expressions are eliminated by successive substitu-
tion of every internal occurrence of f (x) by a new variable y, bounded by existential
quantifier and added to the formula y f (x ) . For example, formula P ( f ( g ( x))) is
replaced by formula y ( y g ( x) P ( f ( y ))) . After all such replacements there will
not be complex functional expressions in the formula. Further, for every attribute
expression f of functional type all its occurrences f ( x1 ),..., f ( xn ) with the different
parameters x1 ,..., xn are considered. Occurrence f ( xi ) is replaced by variable yi ,
bounded by existential quantifier and substitutive equations ( xi x j ) ( yi y j )
are added. Now in the formula there are only simple attributes, and a method consid-
ered in [26] is used.
2.2 Forward Predicate Transformer
In general case, the post-condition looks like ( x, a ) R ( x, a ) C ( x, a ) , where
R ( r1 : t1 r2 : t 2 ...) is a conjunction of assignments and C(x,a) is a formula
part of post-condition.
We will consider three sets of functional expressions (we consider attributes as a
functional expression with 0 arity): r, s and z. Set r (r1 , r2 ,...) consists of the left
parts of assignment, and also of other functional expressions that recursively depend
on the left parts. In other words, r consists of the left parts of assignments and, if
some functional expression f is included into this set, then all functional expressions
in which f occurs are also included in r. Set s ( s1 , s 2 ,...) consists of functional ex-
pressions which have external occurrences (not in arguments of such functional ex-
pressions) in formula part C of post-condition, but do not coincide with expressions
from the set r. Finally, set z ( z1 , z 2 ,...) consists of functional expressions which
have external occurrences in formula D in right parts of assignments and in internal
occurrences (in arguments if functional expressions) of the functional expressions of
formula part C of post-condition and left parts of assignments, but these assignments
are not included in two other sets (including parameter of basic protocol). Now, con-
sidering formulae, from which a post-condition and formula D are constructed as
functions of external occurrences of elements of these sets, we get a presentation of
post-condition in the following form:
B (r , s, z ) (r1 (r , s, z ) : t1 (r , s, z ) r2 (r , s, z ) : t 2 ( r , s, z ) ...) C ( r , s, z ) ,
Predicate transformer is determined by the following formula:
pt ( D (r , s, z ), (r , s, z )) q1 q2 ... ,where
qi (u , v )( D (u , v, z ) R (u , v, z ) Ei (u , v, z ) C ( r , s, z )) ,
R (u , v, z ) (r1 (u , v, z ) t1 (u , v, z )) (r2 (u , v, z ) t 2 (u , v, z )) ...) ,
494 V. Peschanenko, A. Guba and C. Shushpanov
Formula R (u , v, z ) is a quantifier-free part of the assignment formula. Set of the
variables u(v) represents new variables for each attribute expression from r(s) set. The
pt substitutes attributes from r(s) set to variables from u(v) set in corresponded part of
formula.
Each of disjunctive members q i corresponds to one of possible means of identifi-
cation of functional expressions occurring in formulae ( x, a ) , and Ei (u , v, z ) is a set
of equalities and inequalities corresponding to such identification.
To describe the construction of Ei (u , v, z ) we will consider the set M of all pairs
of functional expressions in the form ( f (k ), f (l )), k (k1 , k 2 ,...), l (l1 , l2 ,...) , where
f (k ) is chosen from set z, and f (l ) – from sets r and s. These functional expressions
shall be equal if their arguments were equal before application of basic protocol.
Let’s choose arbitrary subset N M (including an empty set for every pair
( f (k ), f (l )) N we will consider conjunction of equalities
k l , ( k1 l1 k 2 l 2 ...) . We will unite all such conjunctions in one and will add
to it conjunctive negations of all equalities, which correspond to pairs which are not
included into the set N. We will denote the obtained formula as Gi (r , s, z ) . If this
formula is satisfiable, then the choice is successful. Now obviously, f (k ) is not inde-
pendent and shall change the value because Gi (r , s, z ) is true. Thus, f (k ) shall
change the value in the same way as f (l ) . Set Ei (r , s, z ) Gi (r , s, z ) H i ( z , u , v)
where H i ( z , u , v ) is a conjunction of equalities f (k ) w if a variable w corresponds
to f (l ) . Thus, if f (k ) coincides with several functional expressions, it is not impor-
tant what variable is chosen (transitivity of equality) [6].
3 Backward Application of Basic Protocol
Let S (a ) be an environment state after the application of the basic protocol
x( ( x, a ) P ( x, a ) ( x, a )) , where x is parameters of basic protocol, a – attrib-
utes of model, ( x, a ) R ( x, a ) C ( x, a ) , where R (r1 : t1 r2 : t 2 ...) is a
conjunction of assignments and C is a formula part of post-
condition, D ( x, a ) S (a ) C ( x, a ) is a conjunction of environment state and formula
part of postcondition of basic protocol.
3.1 Satisfiability
At first step of application of basic protocol in backward mode satisfiability of con-
junction of environment state and formula part of postcondition of basic protocol is
checked: sat ( D( x, a )) . If the formula is unsatisfiable, then the basic protocol is not
applicable to environment state S (a ) .If not, then process P ( x, a ) is run and after a
backward predicate transformer is applied: pt 1 ( D ( x, a ), ( x, a )) .
Specializations and Symbolic Modeling 495
3.2 Backward Predicate Transformer
A backward predicate transformer considers three sets of functional expressions r, s
and z (as forward too). A postcondition of the basic protocols is represented by the
following formula:
B (r , s, z ) (r1 (r , s, z ) : t1 (r , s, z ) r2 (r , s, z ) : t 2 (r , s, z ) ...) C ( r , s, z )
A backward predicate transformer is determined by the following formula:
pt -1( D(r , s, z ), (r , s, z )) q11 q21 ... , where
qi1 (u, v)( D(u, v, z ) R(u, r , s, z ) Ei (u, v, z )) (r , s, z ) ,
R (u , r , s, z ) (u1 ( r , s, z ) t1 ( r , s, z )) (u 2 t 2 ( r , s, z )) ...), u {u1 , u 2 ,...} ,
Each of disjunctive members qi corresponds to one of possible identification of
functional expressions, occurring in formulae ( x, a ) and environment state S(a),
where Ei (u , v, z ) are sets of equalities and inequalities corresponding to such identifi-
cation. Formula Ei (u , v, z ) is built in the same way as in forward predicate trans-
former [27].
4 Specialization
We propose to use two types of specializations:
1. Specialization by memory usage
2. Specialization by functional symbol
4.1 Specialization by memory usage
Let a1,a2 be sets of attributes from initial environment state and
a1 a2 a1 a2 a , S (a) S1 (a1 ) S 2 (a2 ) is environment state,
B( x, a) x(1 ( x1 , a1 ) 2 ( x2 , a2 ) P( x, a) 1 ( x1, a1 ) 2 ( x2 , a2 )) is basic
protocol, where x1 x 2 x 1 x 2 x .
If B( x, a) x( i ( x, a) P( x, a) ( x, a)) then sat ( S (a) ( i ( x, a)))
i i
sat ( S (a ) i ( x, a )) and pt ( S (a) ( i ( x, a)), ( x, a)) pt ( S (a)
i i i
i ( x, a ), ( x, a )) . So, in the next text we consider basic protocol as B ( x, a ) only.
4.2 Theorem 1
sat ( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ))
sat ( S1 (a1 ) 1 ( x 1 , a1 )) sat ( S 2 (a2 ) 2 ( x 2 , a2 ))
Proving.
Function sat builds closed formula. So,
496 V. Peschanenko, A. Guba and C. Shushpanov
sat ( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ))
(v1 , v2 , x1 , x2 )( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ))
where v1 ,v2 are variables generated for attribute expression which depend on at-
tributes a1 , a2 . It is known that a1 a2 a1 a2 a x1 x 2
x 1 x 2 x . It means that scope of quantifiers could be narrowed:
(v1 , v2 , x1 , x2 )( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ))
(v1 , x1 )( S1 (a1 ) 1 ( x 1 , a1 )) (v2 , x2 )( S 2 (a2 ) 2 ( x 2 , a2 )) i
sat ( S1 (a1 ) 1 ( x 1 , a1 )) sat ( S 2 (a2 ) 2 ( x 2 , a2 ))
Theorem is proved.
This theorem means the following:
1. If S (a ) S1 (a1 ) S 2 (a2 ) and (a, x) 1 ( x1 , a1 ) and S (a ) is satisfiable, then it
is enough to check satisfiability of conjunction of S1 (a1 ) 1 ( x1 , a1 ) for satisfi-
ability checking of S (a ) ( x, a ) . Checking of satisfiability of S 2 (a2 ) is not re-
quired.
2. Checking of each part sat ( Si (ai ) i ( xi , ai )) could be done concurrently.
This case could be easily generalized to a1 ,...., an case, because if it is possible to
build subsets a1i , ai2 ai a1i ai2 a1i ai2 ai and to spilt an environment
state and basic protocol accordingly to the theorem 1, then
sat ( Si (ai ) i ( xi , ai )) sat ( Si (ai ) i ( xi , ai )) . So, after if we say about such
i i
pair of two sets ai , ai ai a1i ai2 a1i ai2 ai , then we understand that it
1 2
could be applicable and for n sets.
Let’s see how forward and backward predicate transformer can be applied.
4.3 Theorem 2
For forward application of basic protocol it is true that:
pt ( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ), 1 ( x 1 , a1 ) 2 ( x 2 , a2 ))
pt ( S1 (a1 ) 1 ( x 1 , a1 ), 1 ( x 1 , a1 )) pt ( S 2 (a2 ) 2 ( x 2 , a2 ), 2 ( x 2 , a2 ))
Proving.
pt function builds sets r , s , z from postcondition 1 ( x 1 , a1 ) 2 ( x 2 , a2 ) and
formula S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ) , where r is a set of attribute ex-
pressions from left parts of assignments of postcondition, s is a set of attribute ex-
pressions from formula part of postcondition, z is a set of other attribute expressions
from formula and postcondition. We know that sets of attribute expressions from pairs
S1 (a1 ) 1 ( x 1 , a1 ) , 1 ( x 1 , a1 ) and S 2 (a2 ) 2 ( x 2 , a2 ) , 2 ( x 2 , a2 ) are not inter-
sected. It means that we could split each set r , s, z on subsets
Specializations and Symbolic Modeling 497
r r1 r 2 , s s 1 s 2 , z z 1 z 2 and r1 r2 , s1 s 2 ,
z 1 z 2 , because a1 a2 . Let’s write formula which is built by pt func-
tion.
Let D(a, x) D1 ( x1 , a1 ) D2 ( x2 , a2 ), D1 ( x 1 , a1 ) S1 (a1 ) 1 ( x 1 , a1 ) ,
D 2 S 2 (a2 ) 2 ( x 2 , a2 ) and 1 ( x 1 , a1 ) R1 (r1 , s1 , z1 ) C1 (r1 , s1 , z1 ) ,
2 ( x2 , a2 ) R2 (r2 , s2 , z 2 ) C2 (r2 , s2 , z 2 ) . So, general formula of predicate trans-
former is the following:
qi (u, v)( D(u , v, z ) R(u, v, z ) ( Ei (u , v, z )) C (r , s, z ))
i i
where R (u , v, z ) R1 (u1 , v1 , z1 ) R2 (u 2 , v2 , z 2 ) , C (r , s, z ) C1 (r1 , s1 , z1 )
C 2 ( r2 , s2 , z 2 ). because (a, x) 1 ( x 1 , a1 ) 2 ( x 2 , a2 ) .
Let’s write in details how to obtain Ei (u , v, z ) . It is known that r1 r 2 ,
i
s 1 s 2 , z 1 z 2 . To build such disjunction we should take into account
all pairs of functional attribute expressions from sets r,s and z. It means that each such
pair should be in set of attribute (r1 s1; z1 ) or (r2 s2 ; z 2 ) . So,
Ei (u , v, z ) ( Ei1 (u1 , v1 , z1 )) ( Ei2 (u 2 , v2 , z 2 ))
i i1 i2
Let’s consider formula of predicate transformer:
pt ( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 1 ( x 2 , a2 ), 1 ( x 1 , a1 ) 2 ( x 2 , a2 ))
qi (u , v)( D(u , v, z ) R (u, v, z ) ( Ei (u , v, z )) C (r , s, z ))
i i
(u1 , u 2 , u1 , v2 )( D1 (u1 , v1 , z1 ) D2 (u1 , v1 , z1 )
Ri (u1 , v1 , z1 ) Ri (u 2 , v2 , z 2 )
( Ei1 (u1 , v1 , z1 )) ( Ei2 (u 2 , v2 , z 2 ))
i1 i2
C1 (r1 , s1 , z1 ) C2 (r2 , s2 , z 2 ))
(u1 , v1 )( D1 (u1 , v1 , z1 ) R(u1 , v1 , z1 ) ( Ei1 (u1 , v1 , z1 ))
i1
C1 (r1 , s1 , z1 )) (u 2 , v2 )( D2 (u 2 , v2 , z 2 ) R(u 2 , v2 , z 2 )
( Ei2 (u 2 , v2 , z 2 )) C2 (r2 , s2 , z 2 )) ...
i2
pt ( D1 ( x 1 , a1 ), 1 ( x 1 , a1 )) pt ( D2 ( x 2 , a2 ), 2 ( x 2 , a2 ))
Theorem is proved.
4.4 Theorem 3
For backward mode it is true that:
498 V. Peschanenko, A. Guba and C. Shushpanov
pt 1 ( S1 (a1 ) C1 (r1 , s1 , z1 ) S 2 (a2 ) C2 (r2 , s2 , z 2 ),
1 ( x 1 , a1 ) 2 ( x 2 , a2 )) pt 1 ( S1 (a1 ) C1 (r1 , s1 , z1 ), 1 ( x 1 , a1 ))
pt ( S 2 (a2 ) C2 (r2 , s2 , z 2 ), 2 ( x 2 , a2 ))
Proving.
R (u , v, z ) R (u1 , v1 , z1 ) R (u 2 , v2 , z 2 ) , C ( r , s, z ) C1 ( r1 , s1 , z1 ) C2 ( r2 , s2 , z 2 )
because (ra , x) 1 (a1 , x 1 ) 2 (a2 , x 2 ) . Ei (u, v, z ) ( Ei1 (u1 , v1 , z1 ))
i i1
( Ei2 (u 2 , v2 , z 2 )) from previous theorem.
i2
pt -1 ( S (a) C (r , s, z ), (r , s, z )) qi1
i
(u , v)( S (a ) C (r , s, z ) R(u , r , s, z ) Ei (u , v, z )) (r , s, z )
i
(u1 , u 2 , v1 , v2 )( S1 (r1 , s1 , z1 ) S 2 (r2 , s2 , z 2 )
C1 (r1 , s1 , z1 ) C2 (r2 , s2 , z 2 )
( Ei1 (u1 , v1 , z1 )) ( Ei2 (u 2 , v2 , z 2 )))
i1 i2
1 (r1 , s1 , z1 ) 2 (r2 , s2 , z 2 )
(u1 , v1 )( S1 (r1 , s1 , z1 ) C1 (r1 , s1 , z1 ) ( Ei1 (u1 , v1 , z1 ))) 1 (r1 , s1 , z1 )
i1
(u 2 , v2 )( S 2 (r2 , s2 , z 2 ) C2 (r2 , s2 , z 2 ) ( Ei2 (u 2 , v2 , z 2 ))) 2 (r2 , s2 , z 2 )
i2
1
pt ( S1 (a1 ) C1 (r1 , s1 , z1 ), 1 ( x 1 , a1 )) pt ( S 2 (a2 ) C2 (r2 , s2 , z 2 ), 2 ( x 2 , a2 ))
Theorem is proved.
Theorem 2 and theorem 3 mean that:
1. Functions pt, pt-1 could be applied separately and concurrently.
2. If postcondition contains 1 ( x1 , a1 ) only, then
pt ( S1 (a1 ) 1 ( x 1 , a1 ) S 2 (a2 ) 2 ( x 2 , a2 ), 1 ( x 1 , a1 ))
S 2 (a2 ) 2 ( x 2 , a2 ) pt ( S1 (a1 ) 1 ( x 1 , a1 ), 1 ( x 1 , a1 ))
pt 1 ( S1 (a1 ) C1 (r1 , s1 , z1 ) S 2 (a2 ) C2 (r2 , s 2 , z 2 ), 1 ( x 1 , a1 ))
S 2 (a2 ) C 2 (r2 , s 2 , z 2 ) pt 1 ( S1 (a1 ) C1 (r1 , s1 , z1 ), 1 ( x 1 , a1 ))
So, functions sat ( Di ( xi , ai )), pt ( Di ( xi , ai ), i ( xi , ai )) are called specialization, be-
cause we could use some special theories for implementation of it.
Specializations and Symbolic Modeling 499
5 Examples of Usage of Specializations
5.1 Examples of Specializations by Memory Usage
Example 1. Concrete values. Let S (a ) (i 2) S (a / i ) be an environment state
where i : int and a / i is a set of all attributes in model except i,
b x((i 0) (i : i 1)) . For application of such basic protocol we should
check satisfiability of the next formula: sat ((i 2) (i 0)) 1 , and the postcondi-
tion should be applied to (i 2) : v((v 2) (v 0) (i v 1)) (i 3) . For such
examples direct C++ translation could be used instead of using some special theories,
and it will work much faster because it doesn’t require any additional checking, just
direct translation into C++ code and compilation of it.
Example 2. Let S (a ) (i 2) S (a / i ) be environment state where i : int and
a / i is a set of all attributes in model except i, b x((i 0) (i : i 1)) . For
application of such basic protocol we should check satisfiability of the next formula:
sat ((i 2) (i 0)) 1 , and the postcondition should be applied to (i 2) :
v((v 2) (v 0) (i v 1)) (i 1) (i 3) . For such examples numerical
intervals could be used. So, S (a ) (i (;2)) S (a / i ) ,
b x((i (0;)) (i : i 1)) . Satisfiability checking looks like just crossing of
two numerical intervals: i (;2) (0;) i (0;2) i [1;1] for integer. Appli-
cation of pt creates the following formula: v((v [1;1]) (i v 1))
i 1 [1;1] i [2;2] . This approach will work faster than general satisfiability
checking and quantifiers eliminations. Such approach could be used for all numeric
and enumerated types.
5.2 Examples of Specializations by Functional Symbol
It is not always possible to represent environment state and basic protocols in the
following way: S (a) S1 (a1 ) S 2 (a2 ) , and B(a, x) x(1 ( x1 , a1 )
2 ( x2 , a2 ) P( x, a) 1 ( x1 , a1 ) 2 ( x2 , a2 ))
where a1 a2 ,
a1 a2 a , x1 x 2 x 1 x 2 x . One of such situation occurs when a value
of functional attribute expression and its parameter has different types and belongs to
the different subsets ai . For example, if functional attribute: i, j : int, f : int T is
defined where T (c1, c2 , c3 ) is enumerated type with three enumerated constants:
c1 , c2 , c3 , then formula ( f (i) c1 ) i 0 could be represented with specializations
as follows: ( f : v1 f (i)) (v1 c1 ) (i 0) . Let b 1 ( f ( j ) : c2 ) be a basic
protocol. Its specialized representation is: b 1 ( f : v1 f ( j )) (v1 : c2 ) 1 .
It is required to merge such data structures for pt function which should consider all
pairs of functional attribute expression from sets r,s and z:
500 V. Peschanenko, A. Guba and C. Shushpanov
( f : v1 f (i )) ( f : v1 f ( j )) ( f : v1 f (i ), v2 f ( j )) . After that basic protocol
should be transformed in the following form:
b 1 ( f : v2 f ( j )) (v2 : c2 ) 1 . It is required to take into account two
possible combinations: (i j ) (i j ) . So, we obtain:
pt (( f : v1 f (i ), v2 f ( j )) (v1 c1 ) i 0, v2 : c2 )
( f : v1 f (i ), v2 f ( j ))
v ((i j ) (v c1 ) (v2 c2 ))
v ((i j ) (v1 c1 ) (v2 c2 )) i 0
( f : v1 f (i ), v2 f ( j )) (v2 c2 ) i 0 (i j )
( f : v1 f (i ), v2 f ( j )) (v1 c1 ) (v2 c2 ) i 0 (i j )
( f : v1 f (i )) (v1 c2 ) i 0 (i j )
( f : v1 f (i ), v2 f ( j )) (v1 c1 ) (v2 c2 ) i 0 (i j )
Let S (a) F ( f1 , f 2 ,..., v1 , v2 , a1 , a2 ) S1 (a1 ) S 2 (a2 ) be an environment state
where f1 f 2 ... are names of functional symbols, v1, v2 are variables for each
functional attribute expression from sets a1 ,a2 correspondently, and
F ( f1 , f 2 ,..., v1 , v2 , a1 , a2 )
( f1 : v11 f1 (t11 , t12 ,...), v12 f1 (t11 , t12 ,...),...,
f 2 : v12 f 2 (t12 , t 22 ,...), v22 f 22 (t12 , t 22 ,...),...,...)
where v11 , v12 a f1 , v12 , v22 a f 2 ,... are variables of type of functional names
f1 , f 2 ,... for each attribute expression, a fi is set of attribute, such as fi a j ,
tij aii {ai } , … - corresponded arguments for each functional with the same name
are in one specialization, and Shostak’s method could be applied for each right part of
equation in F.
Let S (a) F ( f1 , f 2 ,..., v1 , v2 , a1 , a2 ) S1 (a1 ) S 2 (a2 ) . and
b(a) x( Fb ( f1 , f 2 ,..., v1 , v2 , a1 , a2 , x1 , x2 ) 1 (v1 , x1 , a1 )
2 (v2 , x2 , a2 ) P(a, x) 1 (v1.x1 , a1 ) 2 (v2 , x2 , a2 ))
5.3 Theorem 4
sat ( S (a) ( x, a)) sat ( (( f i (ti1 , ti2 ,...) f i (ti1, ti 2 ,...)) (vi k vil ))
(i , k , l )
S1 (v1 , a1 ) 1 (v1 , x1, a1 ) S 2 (v2 , a2 ) 2 (v2 , x2 , a2 ))
sat (qi Si (vi , ai ) i (vi , ai , xi ))
i
where f i (ti1 , ti2 ,...) f i (ti1 , ti 2 ,...) is equality of arguments of functional attribute
expressions.
Proving
Specializations and Symbolic Modeling 501
Let’s define F ( f1 , f 2 ,..., v1 , v2 , a1 , a2 , x1 , x2 ) F ( f1 , f 2 ,..., v11 , v12 , a1 , a2 )
Fb ( f1 , f 2 ,..., v12 , v22 , a1 , a2 , x1 , x2 ) . We combine all equations with the same name of
functional symbol f i and renaming variables names after such union for equations
from basic protocol. After that we obtain sets of variables v1 , v 2 and new basic
proto-
col b(a) x( Fb ( f1 , f 2 ,..., v1 , v 2 , a1 , a 2 , x1 , x 2 ) 1 (v1 , x1 , a1 )
2 (v2 , x2 , a2 ) ... P(a, x) 1 (v1 , x1 , a1 ) 2 (v2 , x2 , a2 )) .
For satisfiable checking we should add corresponded implication for each pair of
equation from F ( f1 , f 2 ,..., v1 , v2 , a1 , a2 , x1 , x2 ) with the same name of functional
symbol f i .
(( f i (ti1 , ti2 ,...) f il (ti1 , ti 2 ,...)) (vi k vil ))
(i , k , l )
(( f i (ti1 , ti2 ,...) f i (ti1 , ti 2 ,...)) (vi k vil ))
i, k , l
Each left and right parts of equation and negation of equations are in the same
specialization. It means that we could build here a disjunction of conjunction. Each
conjunct in such disjunction is qi which will be in one form of our specialization. So,
it means that we could check satisfiability in the following form
sat (qi S i (vi , ai ) i (vi , ai , xi )) .
i
Theorem is proved.
5.4 Theorem 5
pt ( S (a) ( x, a), ( x, a))
( pt ( E1i (v1 , x1 , a1 ), S1 (v1 , a1 ) 1 (v1 , x1 , a1 ), 1 ( x1 , a1 )))
i
pt ( E2i (v2 , x2 , a2 ), S 2 (v2 , a2 ) 2 (v2 , x2 , a2 ), 2 ( x2 , a2 ))))
where
pt ( E ij (vj , x j , a j ), S j (vj , a j ) j (vj , x j , a j ), j ( x j , a j )) qk ,
k
qk (u, v)( S i (u, v, z ) i (u, v, z ) R(u , v, z )
Ei j (u, v, z ) E k (u , v, z ) C (r , s, z )
Proving.
The sets v1 , v 2 are built in the same way as in theorem 3. Let’s consider a general
formula of predicate transformer:
pt ( D(r , s, z ), (r , s, z )) (u , v)( D(u , v, z ) R (u , v, z ) Ei (u , v, z ) C (r , s, z ) .
i
Coefficient Ei (u , v, z ) looks like disjunction of conjunction of all possible
i
matchings with functional attribute expressions from sets r,s and z. So, we can present
502 V. Peschanenko, A. Guba and C. Shushpanov
it as conjunction of two disjunctions: Ei (u , v, z ) ( Ek (u, v, z )) ( El (u, v, z ))
i k l
where Ek (u, v, z ) is disjunction for matching of functional attribute expression
k
where parameters and its value are from different sets of a j . El (u , v, z ) is a dis-
l
junction of matching of other functional attribute expression. Each conjunct of such
disjunction could be considered as a conjunction which depends on different sets of
memory a j . It means that disjunction of conjunction Ek (u, v, z ) could be prepared
k
early before calling of some pt function without corresponded substitution of x,y. So,
Ek (u , v, z ) E1k (v1 , x1 , a1 ) E2k (v2 , x2 , a2 ) . Disjunction El (u , v, z ) could be
k k l
presented in the same way. So, the theorem is proved.
5.5 Theorem 6
pt 1 ( S (a) C (r , s, z ), ( x, a))
( pt 1 ( E1i (v1 , x1 , a1 ), S1 (v1 , a1 ) C1 (v1 , x1 , a1 ), 1 ( x1 , a1 )))
i
pt 1 ( E2i (v2 , x2 , a2 ), S 2 (v2 , a2 ) C2 (v2 , x2 , a2 ), 2 ( x2 , a2 ))))
where
pt 1 ( E ij (vj , x j , a j ), S j (vj , a j ) C j (vj , x j , a j ), j ( x j , a j )) qk ,
k
qk (u, v)( S j (vj , a j ) C j (vj , x j , a j ) R (u , r , s, z )
Ei j (u , v, z ) Ek (u, v, z )) j (r , s, z )
This theorem could be proved in the same mode as theorem 4.
6 Experiments
In this section we present some results from our test suites. All experiments are
devided into several groups. We compare the time of modeling of the satisfiability
and the predicate transformer, presented in the Section 2, and these algorithms with
the specializations.
The first group of experiments refers to specialization by memory usage. Projects
contain formulae in which some attributes have only concrete values. Let us present
one typical real example. This example has a functional attribute of symbolic type
with integer parameters, simple enumerated and simple integer attributes. All of these
integer attributes initialize with concrete values and have concrete values at all times
during trace generation (basic protocols do not change those to symbolic ones). Other
attributes are symbolic. We provide a specialization for attributes, which are always
concrete. The difference of modeling time for this example and for this one special-
ized by concrete values is more than in 3 times. Of course, the speedup depends on
Specializations and Symbolic Modeling 503
project: more concrete attributes we have, more speedup we shall obtain. In [25] it
was shown that speedup could be in thousands times.
The second group of experiments refers also to specialization by memory usage,
but not to concrete values. Examples from this group have enumerated attributes and
integer attributes. Some of the integer attributes memories are intersected, some of
them are independent. First of all, we provide the splitting of formulae into two parts
according to attribute types: enumerated part and integer part. For the enumerated part
we use bitsets, for integer – common Pressburger algorithm. Speedup was about 5-
7%. After we specialize an integer part. We consider the attributes which memory is
independent and obtain speedup in 10 times.
So, the results of comparison of modeling time using general satisfiability func-
tions and functions with specialization are given.
Table 6. Results of experiments
Group of tests General With specializations
algorithm
1 930 sec 300 sec (memory usage/concrete values)
2 300 sec 280 sec (splitting by types)
3 300 sec 33 sec (memory usage/independent memory)
7 Conclusions
Symbolic modeling is a powerful technique for the automated reachability of dead-
locks and violations of user-defined properties. The main complexity of the reachabil-
ity problem is in the complexity of satisfiability and predicate transformer functions.
There are a lot of SMT-based techniques which speed up the satisfiability of formulae
that satisfy some particular theory. We propose a technique that allows to speedup
classical symbolic modeling when formulae could be splitted in several parts and used
some special theories for manipulations with them, which are called specializations.
The mathematical description of the algorithm for constructing specializations is pro-
vided and the correctness of such specializations is proved.
Specializations by memory usage and functional symbols are considered and ex-
amples for each are given.
The nearest plans are the investigation of additional kinds of specialization, be-
cause the more specializations we have, the more speedup we obtain.
References
1. Symbolic Modeling, http://en.wikipedia.org/wiki/Model_checking
2. Letichevsky, A., Gilbert, D.: A Model for Interaction of Agents and Environments. In:
Bert, D., Choppy, C., Moses, P. (eds.) Recent Trends in Algebraic Development Tech-
niques. LNCS 1827, pp. 311–328. Springer Verlag, Berlin Heidelberg (1999)
504 V. Peschanenko, A. Guba and C. Shushpanov
3. Letichevsky, A.: Algebra of Behavior Transformations and its Applications. In:
Kudryavtsev, V. B., Rosenberg, I. G. (eds.) Structural Theory of Automata, Semigroups,
and Universal Algebra, NATO Science Series II. Mathematics, Physics and Chemistry,
vol. 207, pp. 241–272. Springer Verlag, Berlin Heidelberg (2005)
4. Letichevsky A., Kapitonova J., Kotlyarov V., Letichevsky Jr., A., Nikitchenko N., Volkov,
V., Weigert T.: Insertion Modeling in Distributed System Design. Problems of
Programming, (4), 13–39 (2008)
5. Letichevsky, A., Kapitonova, J., Volkov, V., Letichevsky Jr., A., Baranov, S., Kotlyarov,
V., Weigert, T.: System Specification with Basic Protocols. Cybernetics and System
Analysis, (4), 3–21 (2005)
6. Letichevsky, A. A., Godlevsky, A. B., Letichevsky Jr., A. A., Potienko, S. V.,
Peschanenko, V. S.: Properties of Predicate Transformer of VRS System. Cybernetics and
System Analyses, (4), 3–16 ( 2010)
7. Letichevsky, A., Kapitonova, J., Letichevsky Jr., A., Volkov, V., Baranov, S., Kotlyarov,
V., Weigert, T.: Basic Protocols, Message Sequence Charts, and the Verification of Re-
quirements Specifications, In: ISSRE 2004, WITUL (Workshop on Integrated reliability
with Telecommunications and UML Languages) , Rennes, 4 November (2005)
8. Letichevsky, A., Letychevskyi, O., Peschanenko, V.: Insertion Modeling System. In:
Clarke, E.M., Virbitskaite, I., Voronkov, A. (eds.) PSI 2011. LNCS 7162, pp. 262–274,
Springer Verlag, Berlin Heidelberg (2011)
9. Bofill, M., Nieuwenhuis, R., Oliveras, A., Rodríguez-Carbonell, E., Rubio, A.: The Barce-
logic SMT Solver. In: Gupta, Aarti and Malik, Sharad (eds.) CAV 2008. LNCS 5123, pp.
294–298, Springer Verlag, Berlin Heidelberg (2008)
10. Barrett, C., Berezin, S.: CVC Lite: A New Implementation of the Cooperating Validity
Checker. In: Rajeev, A., Peled, D.A. (eds.) CAV '04. LNCS 3114, pp. 515–518, Springer
Verlag, Berlin Heidelberg (2004)
11. Barrett, C., Tinelli, C.: CVC3. In: W. Damm and H. Hermanns (eds.) CAV '07. LNCS
4590, pp. 298–302, Springer Verlag, Berlin Heidelberg (2007)
12. Barrett, C., Conway, C. L., Deters, M., Hadarean, L., Jovanović, D., King, T., Reynolds,
A., Tinelli, C.: CVC4. In: Gopalakrishnan, G., Qadeer, S. (eds.) CAV’11. LNCS 6806,
pp. 171–177, Springer Verlag, Berlin Heidelberg (2011)
13. Cotton, S., Asarin, E., Maler, O., Niebert, P.: Some Progress in Satisfiability Checking for
Difference Logic. In: Proc. FORMATS-FTRTFT (2004)
14. Déharbe, D., Ranise, S.: Bdd-Driven First-Order Satisfiability Procedures (extended ver-
sion). Research report 4630, LORIA (2002)
15. Bozzano, M., Bruttomesso, R., Cimatti, A., Junttila, T., van Rossum, P., Schulz, S., Sebas-
tiani, R.: An Incremental and Layered Procedure for the Satisfiability of Linear Arithmetic
Logic. In: Halbwachs, Lenore (eds.) TACAS’05. LNCS 3440, pp. 317–333, Springer
Verlag, Berlin Heidelberg (2005)
16. Ganai, M. K., Talupur, M., Gupta, A.: SDSAT: Tight Integration of Small Domain Encod-
ing and Lazy Approaches in a Separation Logic Solver. In: H. Hermanns, J. Palsberg.
(eds.) TACAS 2006. LNCS 3920, pp. 135–150. Springer Verlag, Berlin Heidelberg (2006)
17. Audemard, G., Bertoli, P. G., Cimatti, A., Kornilowicz, A., Sebastiani, R.: A SAT based
Approach for Solving Formulas over Boolean and Linear Mathematical Propositions. In:
A. Voronkov (ed.) CADE 2002. LNCS (LNAI) 2392, pp. 195–210. Springer Verlag,
Berlin Heidelberg (2002)
18. Bryant, R.E., Lahiri, S.K., Seshia, S.A.: Modeling and Verifying Systems using a Logic of
Counter Arithmetic with Lambda Expressions and Uninterpreted Functions.. In: Brinksma
Specializations and Symbolic Modeling 505
K., Larsen G. (eds) CAV’04. LNCS 2404, pp. 78–92, Springer Verlag, Berlin Heidelberg
(2002)
19. Dutertre, B., de Moura, L.: A Fast Linear-Arithmetic Solver for DPLL(T). In T. Ball and
R.B. Jones, (eds.) CAV’06. LNCS 4144, pp. 81–94, Springer Verlag, Berlin Heidelberg
(2006)
20. Walther, C., Schweitzer, S.: About veriFun. In: F. Baader (eds.) CADE’03. LNCS 2741,
pp. 322–327, Springer Verlag, Berlin Heidelberg (2003)
21. Ball, T., Cook, B., Lahiri, S.K., Zhang, L.: Zapato: Automatic Theorem Proving for Predi-
cate Abstraction Refinement. In: Alur, R. A., Peled D. A. (eds.) CAV'04. LNCS 3114, pp.
457–461. Springer Verlag, Berlin Heidelberg (2004)
22. de Moura, L., Bjørner, N.: Z3: An Eficient SMT Solver. In: C. R. Ramakrishnan, J. Rehof
(eds.) TACAS'08, LNCS 4963, pp. 337–340. Springer Verlag, Berlin Heidelberg (2004)
23. Barrett, C., de Moura, L., Ranise, S., Stump, A., Tinelli, C.: The SMT-LIB Initiative and
the Rise of SMT. In: Barner S., Harris I. (eds.) HVC 2010. LNCS 6504, pp. 3–3, Springer
Verlag, Berlin Heidelberg (2010)
24. Hutter, F., Hoos, H.H., Leyton-Brown, K., Stuetzle, T.: ParamILS: an Automatic Algo-
rithm Configuration Framework. JAIR, 36, 267–306 (2009)
25. Peschanenko, V. S., Guba, А. А., Shushpanov, C. I.: Mixed Concrete-Symbolic Predicate
Transformer. Bulletin of Taras Shevchenko National University of Kyiv, Series Physics &
Mathematics, 2 (2013) (in press)
26. Barrett, C., Sebastiani, R., Seshia, S., Tinelli, C.: Satisfiability Modulo Theories. Frontiers
in Artificial Intelligence and Applications, 185, 825–885 (2009)
27. Godlevsky, A. B.: Predicate Transformers in the Context of Symbolic Modeling of Transi-
tion Systems. Cybernetics and System Analysis, 4, 91–99 ( 2010)
On a Dynamic Logic for Graph Rewriting ?
Mathias Winckel and Ralph Matthes
Institut de Recherche en Informatique de Toulouse (IRIT)
{winckel, matthes}@irit.fr
Abstract. Initially introduced by P. Balbiani, R. Echahed and A.
Herzig, this dynamic logic is useful to talk about properties on ter-
mgraphs and to characterize transformations on these graphs. Also are
presented the deterministic labelled graphs for which the logical frame-
work is designed.
This logic has been the starting point of a formal development, using the
Coq proof assistant, to design a logical and algorithmic framework useful
for verifyin and proving graph rewriting. The formalization allowed us to
figure out some ambiguities in the involved concepts. This formalization
is not the topic here but the clear view brought to us by the formal work,
so the results will be expressed using the original mathematical objects
of this logic.
Some problems of this logic are demonstrated, relatively to the repre-
sentation of graph rewriting. Some are minor issues but some are far
more important for the adequation between the formulas about graph
rewriting and the actual rewriting systems. Invalidating some resulting
propositions, solutions are given to reestablish the logical characteriza-
tion of graph rewriting, which was the initial purpose.
Keywords. Dynamic Logic, Graph Rewriting, Adequation Issues
Key terms. Formal Method, Model
1 Introduction
Nearly every field of computer science uses graphs to represent data or the
behavior of systems. Then, to get a higher level of dynamics, it uses rewriting
in a more or less formal way to handle manipulation of such objects. In some
fields, as in Formal Methods and Model-Driven Engineering, graphs are one of
the main tools, having advanced methods to express and reason on these graphs
is of great pertinance.
Modal logic allows to express relational properties naturally and, with Kripke
semantics, is closely linked to graphs which are its models. Yet, instead of dis-
cussing graphs and rewriting using hyper-graphs as models, the transformed
?
This work has been funded by the CLIMT project (ANR-11-BS02-016 of the French
Agence Nationale de la Recherche)
On a Dynamic Logic for Graph Rewriting 507
graphs could be directly these models. Dynamic logic as defined by D. Harel
offers by its modalities the possibility to express relations on models when they
have some desired properties. In this spirit, introduced during ICGT 2010, “A
Dynamic Logic for Termgraph Rewriting” [1] was proposed by P. Balbiani, R.
Echahed and A. Herzig as a suitable dynamic logic to describe graphs and
transformations on these graphs.
The termgraph lifts the concept of term to the more general one of graph. It
was initially introduced to represent terms while having a simpler way to talk
about recursion or sharing of sub-terms, what a tree-like structure doesn’t allow
to do easily.
Computer science involves data structures that are usually syntactically rep-
resented as simple terms. It is interesting to develop such a tool for more than
just a graphic representation: using graph rewriting, research could be done on
term rewriting with this rich and powerful layer of language of graph structure.
In the following, in Section 2 we will present the type of graphs we use, then
in Section 3 the dynamic logic for graph rewriting, its syntax and its semantics.
In Section 4 the rewriting system is presented, and propositions to logically talk
about it, but with issues to express these concepts with the logic as originally
introduced. Graph homomorphisms, rewriting steps and application to a graph,
with the definition of rewriting rules, matching of rules and normal forms, actu-
ally leads to some divergences between actual rewriting and its translation using
the logic. Section 4 also discuss and proposes some solutions for these original
issues, for such an utilisation.
2 Termgraph And Rooted Termgraph
To illustrate terms representation as graphs from some typical λ-calculus, as
shown in Figure 1, a classical tree-like term in (a), an example of argument
sharing in (b), and eventually a representation for a fixpoint operator in (c),
usually denoted Y for a function f [2]. One can easily unfold the recursion in
the latter graph and get Y f = f (Y f ), which is what is expected from a
fixpoint operator.
The graphs are deterministic and has labelled nodes and edges [3], but for
convenience edge labels will be named features. A linear graph grammar is de-
fined over a set of nodes N , a set of labels Ω and a set of features F [1], by the
following rules:
N ode ::= n : ω (f1 ⇒ N ode, ... , fk ⇒ N ode) | n : • | n
avec n ∈ N , ω ∈ Ω et f1 , ... , fk ∈ F
T ermGraph ::= N ode | N ode + T ermGraph
Allowing one to define a node, with its label and its direct sons, a node
without label or just a reference to an actual node with the first rules, and to
define multiparts termgraphs with the others.
508 M. Winckel and R. Matthes
app app
app
λx y app
f
x F x
(a) term: (λx.x) y (b) sharing: F x x (c) recursor: Y f
Fig. 1. Examples of term representation
In a less syntactic and maybe more semantic manner, a termgraph is defined
as well by a structure.
(N , E, LN , LE , S, T ) with
– N a finite set of nodes
– E a finite set of edges
– LN a partial function from N to Ω, associating a node to its label
– LE a total function from E to F, associating a edge to its feature
– S a source function from E to N
– T a target function from E to N
It’s assumed that the previous definitions respect a determinism condition de-
fined as
∀ e1 , e2 ∈ E, S(e1 ) = S(e2 ) ∧ LE (e1 ) = LE (e2 ) → e1 = e2
Rooted TermGraphs. In the following, for the need of the logic, the graphs
will be an extension of the termgrah with a specific node pointed as its root.
(N , E, LN , LE , S, T , r) with the root r ∈ N
3 Dynamic Logic for Graph Rewriting
A modal logic is a propositional logic, extended with one or several modality op-
erators. Dynamic logic [4] is a multi-modal logic, its modalities being actions and
the possibility to express a choice, an iteration or the sequence of actions. Actions
can be defined to express graph transformations by miscellaneous modalities.
3.1 Syntax of the Dynamic Logic
For given countable sets F and Ω, of features and labels (their respective ele-
ments being usually denoted a, b, ... and ω, π, ...), the rules defining the formulas
and the actions of the syntax are the following. [1]
For an action α:
On a Dynamic Logic for Graph Rewriting 509
α ::= a for the navigation by a ∈ F from the root
|U for changing the root to any node in the graph
|n|n for the creation of a node n, eventually setting it as the root
| φ? for the verification of the validity of a formula φ for the root
| (ω :=l φ) | (ω :=g φ) for labeling a node with a label ω ∈ Ω if a
formula φ is valid, locally for the root or globally for any node.
| (a + (φ, ψ)) | (a − (φ, ψ)) for adding or removing edges with the
feature a, between nodes verifying a formula φ and those verifying a formula
ψ.
| (α1 ; α2 ) | (α1 ∪ α2 ) | α1∗ for the definition of a sequence, a choice
or an iteration over actions α1 and α2 .
For a formula φ:
φ ::= ω | ⊥ | ¬ φ | φ ∨ ψ | [ α ]φ
ω as a formula means that node labels can be atomic formulas. And intuitively,
[ α ]φ means that after any execution of an action α, the formula φ holds. The
propositional logic of such dynamic logic being a classical one, some more sym-
bols can be defined as usual conjunction, implication and equivalency being re-
spectively φ∧ψ ≡ ¬(¬φ∨¬ψ), φ → ψ ≡ ¬φ∨ψ and φ ↔ ψ ≡ (φ → ψ)∧(ψ → φ),
for given formulas φ and ψ. One more modality can be defined, for a given action
α and formula φ, as < α > φ ≡ ¬[α]¬φ. Intuitively, it holds when an execution
of the action α makes φ hold.
3.2 Semantics of the Dynamic Logic
A semantic can be defined for this dynamic logic using rooted termgraphs as
models and verifying the properties expressed by the formulas on them, using
the sets of labels and features of the logic to define them. [1]
Models. The rooted termgraphs used here are different from the previous ones.
The labeling function for nodes is now defined over the power set of Ω, so typed
LN : N → P(Ω), of what the former definition is said to be a particular case.
Interpretation of Actions and Formulas. Before the definition of a satisfi-
ability relation, it needs interpretation functions over the models.
IG is defined as
– IG (a) = {e | e ∈ E and LE (e) = a} the set of a edges.
– IG (ω) = {n | n ∈ N and ω ∈ LN (n)} the set of nodes having the ω label.
And, for any a ∈ F, RG is a family of binary relations defined such as
– RG (a) = {(n1 , n2 ) | ∃e ∈ IG (a), S(e) = n1 and T (e) = n2 }
510 M. Winckel and R. Matthes
Because of dependency between formulas and actions, the satisfiability rela-
tion |= for a formula F and a rooted termgraph G requires an inductive definition
on F , dependent of a relation G −→α G0 for every action α.
– G ω iff n0 ∈ IG (ω), interpreting ω in G.
– G 2⊥
– G ¬φ iff G 2 φ
– G φ ∨ ψ iff G φ or G ψ
– G [α]φ iff for any rooted termgraph G0 , if G −→α G0 then G0 φ
with G −→α G0 the binary relation between two termgraphs G and G0 consid-
ering the action α. It is defined inductively on α, with G = (N , E, LN , LE , S, T , r)
0 0
and G0 = (N 0 , E 0 , LN , LE , S 0 , T 0 , r0 ), but only definitions for the useful cases
will be introduced. The definitions are declarative, so it should be read as condi-
tions for a correct relation between G and G0 and not as the way to get a model
G0 from a model G.
A notation JeKG is introduced for a graph G and an edge e of G, to express
(S(e), LE (e), T (e)). Ambiguity decreases, in the following definitions, the set E
being only identifiers of edges for the functions LE , S and T and not tuples of
these informations. We can justify this by looking at the other way, with E as a
set of tuples, and seeing that it does not make much sense: for example, when
redirecting edges, the set E was staying the same, and so were the tuples in this
set, but the target of an edge was changed and thus a tuple was associated by
functions to information which is no longer the content of the tuples.
For convenience, another notation G[n0 ] is introduced for a node n0 of a graph
G = (N , E, LN , LE , S, T , r), to express the graph (N , E, LN , LE , S, T , n0 ), or in
more simple terms, to express changing the root of G with n0 .
G −→a G0 iff
0 0
– N 0 = N , E 0 = E, LN = LN , LE = LE , S 0 = S and T 0 = T , to express
almost all parts of G0 being the same.
– (n0 , n00 ) ∈ RG (a), to express the possibility to navigate from the root of G
to the root of G0 .
G −→U G0 iff
0 0
– N 0 = N , E 0 = E, LN = LN , LE = LE , S 0 = S and T 0 = T , to express no
characterization for the root of G0 but other parts being the same.
G −→(ω:=g φ) G0 iff
0
– N 0 = N , E 0 = E, LE = LE , S 0 = S, T 0 = T and r0 = r
0 0
– for any m ∈ N , if G[m] φ then LN (m) = LN (m) ∪ {ω} else LN (m) =
N
L (m)\{ω}, expressing the addition or deletion of the label ω of any node
m of the graph satisfying the formula φ.
G −→φ? G0 iff
On a Dynamic Logic for Graph Rewriting 511
0 0
– N 0 = N , E 0 = E, LN = LN , LE = LE , S 0 = S, T 0 = T and r0 = r
– G φ, to express the formula φ being valid for G.
G −→(ω:=l φ) G0 iff
E0
– N 0 = N , E 0 = E, L
0
= LE , S 0 = S, T 0 = T and r0 = r
N0
– if G φ then L (r) = L (r) ∪ {ω} else L (r) = LN (r)\{ω}, to express
N N
the addition or deletion of the label ω from the root.
G −→(a+(φ,ψ)) G0 iff
0
– N 0 = N and LN = LN
Considering the set of candidate edges C = {(ns , a, nt ) | with ns ∈ N and nt ∈
N such as G[ns ] φ and G[nt ] ψ}, to be added only between nodes vali-
dating the formulas φ and ψ.
– E 0 ⊃ E, and for all e ∈ E, JeKG0 = JeKG , characterizing an addition without
any loss.
– for all p ∈ C, ∃e ∈ E 0 , JeKG0 = p and for all e ∈ E 0 \ E, JeKG0 ∈ C, expressing
candidate edges being added in E 0 but nothing else.
G −→(a−(φ,ψ)) G0 iff
0
– N 0 = N and LN = LN
Considering the set of deleted edges E = {e | e ∈ E such as JeKG0 = (ns , a, nt )
with ns and nt such as G[ns ] φ and G[nt ] ψ}.
– E 0 = E \ E characterizing deletion of edges only between nodes validating
formulas φ and ψ.0
– for all e ∈ E, LE (e) = LE (e), S 0 (e) = S(e), T 0 (e) = T (e) and r = r0 .
G −→α;β G0 iff
– there exists a rooted termgraph G00 such G −→α G00 and G00 −→β G0 .
G −→α∗ G0 iff
– there is a rooted termgraph sequence (G(0) ,. . . ,G(k) ) with G(0) = G, G(k) =
G0 and for all i ∈ {0, ... , k − 1}, G(i) −→α G(i+1) .
Semantics of left over actions n, n and α1 ∪ α2 are in the original paper.
At this point, the logic allows to characterize classes of graphs using the satisfi-
ability of formulas by these graphs as models. Everything goes pretty well, but
issues come when dealing with rewriting and propositions made to talk about
graph rewriting.
4 Actual Rewriting, and its Issues
The approach here is an algorithmic one: transformation actions are defined
within a rewriting system and can be applied to a graph, alone or sequentially.
It forms rules of rewriting that could be applied if an instantiation of the rule
is found in a given graph. Such instance can be defined with a graph morphism
which embeds the graph domain of the rule into the graph in which the rule
could be applied.
512 M. Winckel and R. Matthes
4.1 Homomorphism of Graphs
A homomorphism of labelled graphs h : G → G0 can be defined, given two rooted
0 0
termgraphs G = (N , E, LN , LE , S, T , r) and G0 = (N 0 , E 0 , LN , LE , S 0 , T 0 , r0 ).
n 0
Somewhat, only with a function h : N → N preserving the labeling of nodes
but equally preserving the source and target function for the edges, and thus the
labelization of edges.
0
So ∀m ∈ N , LN (hn (m)) = LN (m) is mandatory and then because of the deter-
minism condition satisfied by the graphs, there is no ambiguity on the conditions
for the edges of the codomain graph G0 . For any e of E, it only requires one exist-
0
ing e0 of E 0 verifying S 0 (e0 ) = hn (S(e)), LE (e0 ) = LE (e) and T 0 (e0 ) = hn (T (e)),
and so with corresponding source and target while preserving the edge labeliza-
tion, mandatory too for such homomorphism. There is no specific condition for
correspondence of roots of the two termgraphs.
Examples of the original paper can be presented here, in Figure 2, to display
some homomorphisms: morphisms h2 and h3 , between three graphs B1, B2 and
B3 displaying the association of their nodes.
n0 : f n0 : f n0 : f
h2 : B1 −→ B3 h3 : B2 −→ B3
1 1 1 n0 7→ n0 n0 7→ n0
n1 : g n1 : g n1 : g n1 7→ n1 n1 7→ n1
a b a b a b n2 7→ n2 n2 7→ n2
n2 : • n3 : • n2 : 0 n3 : • n2 : 0 n3 7→ n2 n3 7→ n2
B1 B2 B3
(a) termgraphs (b) homomorphisms
Fig. 2. Examples of termgraph homomorphisms
Existence of Graph Homomorphisms and Particularity of such
Homomorphisms
In the original paper, a way to talk about homomorphisms on graphs using
the logic is proposed, a formula express this concept and is defined for a graph
G = (N , E, LN , LE , S, T , r). Thus an action αG and a formula φG can relate that
there is a homomorphism from G to a graph G0 if and only if G0 < αG > φG .[1]
For this, considering N = {n0 , . . . , nN −1 }, with N being the number of nodes
of G and n0 being its root, and considering a sequence P = {π0 , . . . , πN −1 } of
distinct elements of Ω, each πi is going identify the node ni .
The action αG is defined
– with βG = (π0 :=g ⊥) ; . . . ; (πi :=g ⊥) ; . . . ; , (πN −1 :=g ⊥), characterizing
the elimination of any label πi .
On a Dynamic Logic for Graph Rewriting 513
i
– with, for 0 ≤ i ≤ N − 1, γG = (¬π0 ∧ . . . ∧ ¬πi )? ; (πi :=l >) ; U , charac-
terizing the labelization with πi of a node not already labelled with πk for
k ≤ i.
0 N −1
– and finally αG = βG ; γG ; . . . ; γG , sequencing these actions.
Then, the formula φG is defined
– with 0 ≤ i ≤ N − 1, if LN (ni ) 6= ∅ then ψG i
=< U > (πi ∧ LN (ni ))
i
else ψG = >, characterizing the conservation of the labelization of a node
identified as the image of ni , by the label πi , but nothing if this node wasn’t
labelled.
– with for 0 ≤ i, j ≤ N − 1, if ∃e ∈ E such S(e) = ni and T (e) = nj then
ζ i,j E i,j
G =< U > (πi ∧ < L (e) > πj ) else ζG = >, characterizing the existence
of edges corresponding to the ones of G.
0 N −1 0,0 N −1,N −1
– and finally φG = ψG ∧ . . . ∧ ψG ∧ ζG ∧ . . . ∧ ζG , verifying all these
formulas.
In the latter definition, it is assumed to have a subset P of Ω of fresh labels not
already used in G, and so that do not have to be preserved by the homomorphism
in G0 , and do not erase information when used in βG . Such dedicated labels for
identification of the elements of N could be assumed as a part of the set Ω, by
definition. This definition of P , identifying differently each node of G and the
way it is used in the formula, requires more for a homomorphism than what
implies the homomorphism definition. Acutally, the examples of the Figure 2,
which do not stand against the definition, are not injective morphisms. In the
formula < αG > φG , matched nodes are labelled to be identified to only one
node of G, as the test (¬π0 ∧ . . . ∧ ¬πi )? express it. Therefore, it is mandatory
for a homomorphism h : G −→ G0 to be injective to satisfy the formula and
this necessary condition on the models. Although, such injectivity is a common
feature of graph morphism in many graph rewriting frameworks, so it is not
mandatory to define formally such a system using injective matching only and the
initial definition of graph homomorphism should specify wether this particularity
is actually required.
4.2 Rewriting Step and Translation in Logic
The rewriting is defined to be applied to a graph G = (N , E, LN , LE , S, T , r)
0 0
to obtain a graph G0 = (N 0 , E 0 , LN , LE , S 0 , T 0 , r0 ), the result of the application
of an action α to a graph G is denoted α[G]. It is possible to make sequences
of actions, as empty or the concatenation of an action α and another sequence,
and the application of a sequence ∆ to a graph G is denoted ∆[G].[1]
– if ∆ is the empty sequence then ∆[G] = G
– if ∆ = α; ∆0 is a concatenation, with a sequential operator “;”, then ∆[G] =
∆0 [ α[G] ]
For a morphism h and a sequence ∆, h(∆) denotes the sequence one obtains by
substitution in ∆ of any node n by h(n).
514 M. Winckel and R. Matthes
The original paper proposes a way to talk about sequences of actions, by de-
scribing the sequence of rewriting actions as a sequence of logical actions. Thus,
after the translation of an action α of the rewriting system, the relation −→tr(α)
introduced with the semantic of the logic allows to relate models, the second
potentially being the result of the application of the action to the first one.
The translation of a sequence of a given action a and another sequence ∆ will
be the sequence of translations αa;∆ = αa ; α∆ . Assuming that any action has
a translation entirely independent of the rest of the sequence, the translation
order actually does not matter, but a sequential translation seizes a correct idea.
For the rewriting actions:
n a m is a local redirection.
An outgoing edge from a node n with the feature a is modified to point to
a node m.
0 0
– N 0 = N , E 0 = E, LN = LN , LE = LE , S 0 = S and r = r0 , the nodes, edges,
their labelization, the source of the edges and the root are the same.
0
– for e ∈ E such S 0 (e) = n and LE (e) = a then T 0 (e) = m, the target of the
only wanted edge is changed for m. ∀e0 ∈ E 0 , if e0 6= e then T 0 (e0 ) = T (e0 ),
the target function doesn’t change for the other edges.
To begin with, here is the given formula for the local redirection.
– αn a m = (a − (πn , >)) ; (a + (πn , πm ))
One can notice that this formula depicts a transformation by deleting an
edge, labelled a, between a node n and any other node, though the determinism
dictates the existence of at most only one such edge. But above all thus an a
edge between this node n and a node m is added at the end. Looking at the
definition of the local redirection, one can see that E 0 = E specify that no edge
is actually added and the other item clearly specify that only an existing e of E
with source as n and feature as a has its target changed to m.
A difference happens between the rewriting of a graph and the models linked
by the actions of the translation of the rewriting action, as shown in Figure 3
which is a counter example for adequation between this translation and rewriting.
n : πn n : πn n : [πn ] n : [πn ]
=⇒ −→ a
m : πm m : πm m : [πm ] m : [πm ]
(a) rewrite rule (b) models
Fig. 3. Local Redirection Difference
On a Dynamic Logic for Graph Rewriting 515
The problem being that the logical actions don’t require the existence of the
redirected edge, what is obviously mandatory to rewrite it. We proposed, as a
correction for this problem, the formula:
– αn a m = (λn :=g ⊥) ; (λn :=g πn ∧ < a > >) ; (a−(πn , >)) ; (a+(λn , πm ))
The added actions require the marking of a node n with a λn , assumed fresh
for the graph, validating the possibility to navigate along a a edge to any node,
then it proceeds to delete and add the edge only with this sole mark. If no edge
was removed, no mark will be found on the node n, correcting this problem of
adequation between rewriting and logical translation.
n m is a global redirection.
The target of every edge pointing to the node n is modified to make the
edges point to the node m.
0 00 0
– N 0 = N , LE = E, LN = LN , LE = LE and S 0 = S, the nodes, edges, their
labelizations and the source of the edges.
– for all e ∈ E such as T (e) = n then T 0 (e) = m else T 0 (e) = T (e), the targets
of the only wanted edges are changed for m, otherwise it does not change.
– if n = r then r0 = m else r0 = r, the root changes if it’s the former target of
the changed edges.
If one looks at the translation of the global redirection, with first a formula
to globally redirect to a node m every a edges initially pointing to a node n:
– αn ga m = (λa :=g ⊥) ; (λa :=g < a > πn ) ; (a − (>, πn )) ; (a + (λa , πm ))
This formula does not suffer of this problem, the second action marking with a
λa any reachable node by an edge with the feature a. Then, the action adding
the edges does not add wrongly an inexistant previously removed edge, since the
label λa ensures these nodes are subjects to redirection. However, it is implicitly
required that this λa is a dedicated label for this action, to not interfere with
any other formula using this label for another purpose. The full translation of
the rewriting step is finally a sequence of the previously defined formulas, for
every feature of the graph:
– αn m = ;a∈F αn ga m
Comes finally the last rewriting action, the node labelization.
n : ω (f1 ⇒ n1 , ... , fk ⇒ nk ) is a node definition or labelization.
It adds a node n, if it does not already belongs to the graph, or modifies
an already existing one. It assigns the label ω and defines the edges e1 to ek
outing of this node n, respectively pointing to the nodes n1 to nk with the
labels f1 to fk , according to the following definition:
– N 0 = N ∪ {n, n1 , ..., nk }, nodes of the rules which are not already included
in G are added.
516 M. Winckel and R. Matthes
00 0
– LN (n) = ω and ∀m ∈ N \ {n}, LN (m) = LN (m), n is labelled with ω and
the other nodes keep the same labeling.
0
Considering the newly defined edges E = {ei |S 0 (ei ) = n, LE (ei ) = fi and
T 0 (ei ) = ni } with 1 ≤ i ≤ k.
– E 0 = E ∪ E, new edges are added to the already existing ones.
0
– ∀ei ∈ E, LE (ei ) = fi , the features of the new edges are defined, and ∀e ∈ /
E0
E, L (e) = LE (e), the features of the other edges don’t change.
– ∀ei ∈ E, S 0 (ei ) = n and ∀e ∈/ E, S 0 (e) = S(e), the same thing is done for the
source function.
– ∀ei ∈ E, T 0 (ei ) = ni and ∀e ∈ / E, T 0 (e) = T (e), the same thing is done for
the target function.
– r0 = r, the root remains the same.
And the formula, initially given as its translation:
– αn : ω (f1 ⇒n1 , ... , fk ⇒nk ) =
U ; πn ? ; (ω :=l >) ; (f1 + (πn, , πn1 )) ; ... ; (fk + (πn, , πnk ))
One could there raise a first issue, regarding the end of the formula: it adds
without much care outgoing edges of the node n, thereby not ensuring the deter-
minism. The result is that no model can be related as resulting of the application
of this action, when what was intended was a relation on models displaying a
redirection of these edges. We proposed this correction, using as previously the
idea of edge redirection by deletion of the former edge and addition of the redi-
rected one.
– αn : ω (f1 ⇒n1 , ... , fk ⇒nk ) =
U ; πn ? ; (ω :=l >) ;
(f1 − (πn , >)) ; . . . ; (fk − (πn , >)) ; (f1 + (πn , πn1 )) ; . . . ; (fk + (πn , πnk ))
Now, a model related by the relation of the corresponding formula for this
action can exist, and it will be a deterministic graph with redirected edges. How-
ever, there still remains a problem. If one looks at the transformation expressed
by the formula, it translates the labelization of a node positionning the root
on a node n, identified by πn , thus it labelizes it with ω and finally changes the
edges outgoing of this node n. As displayed in Figure 4, this raises again a major
difference between the graph one obtains by rewriting and the models related by
the relation for the modality of the logical action. The behavior is not the same
because the formula requires only the addition of a label to the multiple already
existing labels of a node while the rewriting action requires the replacement as
an unique label.
For a better understanding of the extent of this problem, one should look at
the following, which demonstrates how much of an issue that becomes in regard
of characterization of rewriting system.
On a Dynamic Logic for Graph Rewriting 517
n:l =⇒ n:ω n : [l] −→ n : [ω, l]
(a) rewrite rule (b) models
Fig. 4. Node Labeling Difference
4.3 Rewriting Rules and Rewriting System Characterization
Rewriting rules
A rewriting rule is expressed as a graph L and a sequence of actions ∆ to apply,
and will be noted (L, ∆). It is said that a graph G is rewritten as a graph G0
if there exists a homomorphism h : L → G such as h(∆)[G] = G0 , denoted
G →L,∆ G0 .
A simple yet useful example may be defined, representing an action on a simple
on-off switch of an electrical network:
L= off : SwitchOff
e
on off
pos : • e e neg : •
int : Off on : SwitchOn
Fig. 5. Left-hand side L of a rule
Displayed in Figure 5, on the left of L there is a positive terminal of the
circuit, on the right side there is a negative one. The edges labelled edges as e are
for the electrical circuit and the switch is currently set to Off . The two positions
SwitchOn and SwitchOff are linked to each other to avoid any mismatch if
another switch is part of the network, even though a one-way link alone could
be enough.
∆switchOn = (int : On ( ) ; int >>e on)
Fig. 6. Sequence ∆ of a rule
The sequence for this rule, as displayed in Figure 6, is made of an action to
mark the switch node with On without adding any extra edge, and an action to
redirect the edge e, outgoing from this node to the Off position of the switch, to
get a graph representing a closed electrical circuit, displayed now in Figure 7 as
the result on the graph L.
518 M. Winckel and R. Matthes
off : SwitchOff
on off
pos : • e e e neg : •
=⇒ int : On on : SwitchOn
Fig. 7. result of applying ∆ to L
Rewriting System Characterization and The Labelization Problem
Using the previous example of a switch activation, a graph G is displayed in
the Figure 8, being a simple electrical network having only a generator and a
switch, and being really similar to the left-hand side of the rule (L, ∆switchOn )
previously defined.
off : SwitchOff
e
on off
e e
G = pos : DC+ int : Off on : SwitchOn neg : DC−
Fig. 8. graph G
The normal form of the graph G, with respect to the rule (L, ∆switchOn ),
satisfies the formula ψ := [U ]¬Off , because of its switch being activated and
thus being labelled with On, so no node is labelled with Off anymore.
In the end of the original paper, the proposition 4 [1] which defines a way to
talk about a normal formal of a graph G satisfying ψ says that this should be
equivalent to the following satisfaction
G [(αL ; φL ?; α∆switchOn )∗ ]([αL ; φL ?] ⊥ → ψ)
To explain this proposition, one can think about a normal form with respect
of a rule as resulting of successive matchings and applications of this rule. When
the rule cannot be matched anymore this implies that the formula ψ holds, which
characterizes the normal forms. In the example with the rule (L, ∆switchOn ),
every switch must be activated and the rule shouldn’t be matchable then, because
there is no label Off or any e edge pointing to the right position. But because of
the difference between rewriting and translation in the logic, when every e edge
of the switch is redirected in the models, no more matching is available but there
is still the label Off , along with the label On labeling the node int. The formula
ψ is thus unsatisfied, the graph G and the formula ψ are a counter example for
this proposition.
On a Dynamic Logic for Graph Rewriting 519
First, one can notice that relying on the definition of the models, the logic
makes use of the labels to mark and reason, but without any difference between
the ones being actual parts of the graph and the ones serving the logic. Even if
the problem appears with the semantics, the solution is not there, because any
change of the model definition, allowing to make a difference between graph and
logic for example by simply splitting in two layer these informations, doesn’t
mean that the same syntax of the logic allows to express this difference.
The problem comes with the translation of the label definition of a node, it
seems sounding to complete this translation to get the correspondence between
rewriting and relation on models. But the rules and the matching use the labels
to logically identify the node, so one should be careful with erasing, thus to erase
everything is not an option here. Actually, the syntax of the logic needs explicitly
the label to delete, with a global or local labeling action. And the syntax of the
action currently does not allow to say which label will be erased. This is where
the problem lies in: there is only the identifier of a node and the new label.
To actually translate the action, it is possible if it is given more information, so
a new label definition action is a possibility or the syntax of the logic should be
changed to express actions on a new model type.
But before any syntactic change, of rewriting actions or logic, it is interesting
to consider the action in the context of a graph. Because of the LN : N −→ Ω
of the graph, and having already the node n of the action, the labeling function
can give the label to remove. Regarding of the action to translate, there is not
this information, but more generally rewriting actions are defined for a specific
graph. Thus, lifting the translation in formula to the level of the rule, this gives
the required information. From this rule, a graph is given as left-hand side and
thus a LN : N −→ Ω function can be used, and it allows an explicit erasing of
a label as it was implicitly expected by the rewriting action.
For a graph G = (N , E, LN , LE , S, T , r) and a node labeling action n :
ω (f1 ⇒ n1 , ... , fk ⇒ nk ), we give as a correct translation the formula
– αG, n : ω (f1 ⇒n1 , ... , fk ⇒nk ) =
U ; πn ? ; (LN (n) :=l ⊥) ; (ω :=l >) ;
(f1 − (πn , >)) ; ... ; (fk − (πn , >)) ; (f1 + (πn , πn1 )) ; ... ; (fk + (πn , πnk ))
However, the application of the rule to the left-hand side results in another
graph with another labeling function and the translation of a sequence cannot
remain as previously proposed. Every node labeling action should be translated
depending on the result of previously applied actions. The translation of a se-
quence of an action a and another sequence ∆ is a sequence of sequential trans-
lations:
αG, (a;∆) = αG, a ; αa[G], ∆
One could notice that this translation is still done in a linear way through the
sequence, but this definition could actually be slightly changed because every
action but node labelization can be translated independently. Only the node
labelization is dependent of a context and more precisely there are conflicts only
520 M. Winckel and R. Matthes
when editing the same node, having to erase a label which was just defined
previously in the sequence.
5 Conclusion
This paper provided an introduction to a dynamic logic, defined by P. Balbiani,
R. Echahed and A. Herzig. Originally, a formalization was done using the
Coq proof assistant and so was done a study of this logic. During this work
some mistakes were spotted. While sometimes these were just minor-looking
imprecisions, when working with formal logic, it may be relevant to point out
such ambiguities if the logic can become incoherent with a wrong interpretation.
Other mistakes are not only due to interpretation, and as it was demon-
strated, they allow to find counter-examples for propositions and thus to estab-
lish incoherence with the main goal of this logic, to talk about rewriting systems.
Solutions to these issues are proposed, they resolve the issues by getting back to
an adequation between actual rewriting and relations on models of the logic.
A first idea to find another technical problem is the conservation of deter-
minism of graphs, which only seems currently assumed and can be broken by
the action of the logic, heading to the impossibility to relate a model to another
because of the determinism, assumed by definition. It could be more explicit in
the semantics of rewriting, and currently the definition of one of the rewriting
actions allows to add edges breaking this condition. This is something the logic
can express and handle during the translation of these rewriting actions, this
looks like an interesting improvement of the logical framework.
The models are defined with a multi-labeling function of nodes and already
demonstrate some differences during translation of rewriting into the logic, be-
cause the models are graphs with a unique-labeling function into the initial
rewriting system. It is not totally clear whether the difference stops there, and
it is required to study more uses of the logic in regards of rewriting to be sure
of the correct use. It seems interesting to study the reverse translation as well,
from logic to rewriting, to get an useful and complete logical framework, in the
idea of Curry − Howard correspondence with terms as proofs.
References
1. Balbiani, P., Echahed, R., Herzig, A.: A dynamic logic for termgraph rewriting. In
Ehrig, H., Rensink, A., Rozenberg, G., Schürr, A. (eds.) ICGT. LNCS, vol. 6372,
pp.59–74. Springer, Heidelberg (2010)
2. Ariola, Z.M., Klop, J.W.: Lambda calculus with explicit recursion. Inf. Comput.
147(2), 154–233 (1997)
3. Barendregt, H., van Eekelen, M., Glauert, J., Kennaway, J., Plasmeijer, M., Sleep,
M.: Term graph rewriting. In de Bakker, J., Nijman, A., Treleaven, P. (eds.)
PARLE Parallel Architectures and Languages Europe. LNCS, vol. 259, pp. 141–
158. Springer, Heidelberg (1987)
4. Harel, D., Kozen, D., Tiuryn, J.: Dynamic logic. Handbook of Philosophical Logic,
MIT Press. 497–604 (1984)
Logical Foundations for Reasoning about
Transformations of Knowledge Bases? ??
Mohamed Chaabani1 , Rachid Echahed2 and Martin Strecker3
1
LIMOSE, University of Boumerdès, Algeria
2
Laboratoire d’Informatique de Grenoble
http://membres-liglab.imag.fr/echahed/
3
Université de Toulouse / IRIT
http://www.irit.fr/~Martin.Strecker/
Abstract. This paper is about transformations of knowledge bases with
the aid of an imperative programming language which is non-standard in
the sense that it features conditions (in loops and selection statements)
that are description logic (DL) formulas, and a non-deterministic assign-
ment statement (a choice operator given by a DL formula). We sketch an
operational semantics of the proposed programming language and then
develop a matching Hoare calculus whose pre- and post-conditions are
again DL formulas. A major difficulty resides in showing that the for-
mulas generated when calculating weakest preconditions remain within
the chosen DL fragment. In particular, this concerns substitutions whose
result is not directly representable. We therefore explicitly add substitu-
tion as a constructor of the logic and show how it can be eliminated by
an interleaving with the rules of a traditional tableau calculus.
Keywords. Description Logic, Graph Transformation, Programming Lan-
guage Semantics, Tableau Calculus
Key terms. MathematicalModel, SoftwareSystem, KnowledgeRepresen-
tation
1 Introduction
Knowledge bases (KBs) are specific forms of graphs structures that are subject
to change because the world they describe changes. The question explored by
this paper is: What is an adequate formalism for describing these changes, and
how to reason about the effects of changes?
Reasoning about graph transformations in full generality is hard [7]. Some
decidable logics for graph transductions are known, such as MSO [6], but are
?
Preliminary workshop version of a paper to be presented at DL 2013
??
Part of this research has been supported by the Climt project (ANR-11-BS02-016).
522 M. Chaabani, R. Echahed and M. Strecker
descriptive, applicable to a limited number of graphs and often do not match with
an algorithmic notion of transformation. Some implementations of verification
environments for pointer manipulating programs exist [9], but they often impose
severe restrictions on the kind of graphs that can be manipulated, such as having
a clearly identified spanning tree.
In [4], the have authors introduced a dynamic logic which is very expressive.
It has been designed to describe different kinds of elementary knowledge bases
transformations (addition of new items, addition and deletion of links, etc.). It
allows also to specify advanced properties on graph structures which go beyond
mu-calculus or MSO logics. Unfortunately, the expressive power of that logic has
a price: the undecidability of the logic. The purpose of the present paper is to
identify a programming language together with a logic such that the transfor-
mation of the KB is decidable. The transformations themselves are not encoded
in the logic itself (as in [4]) but in a dedicated imperative language for which we
develop a Hoare-style calculus.
Work on (KB) updates [8] seem to approach the problem from the opposite
direction: Add facts to a KB and transform the KB at the same time such that
certain formulas remain satisfied. In our approach, the KB the modification of
the KB is exclusively specified by the program.
The work described in this paper is ongoing, some results are still prelimi-
nary. Based on previous work [5], we are in the process of coding the formalism
described here in the Isabelle proof assistant [10]. Parts of the coding in this
paper are inspired by formalizations in the Isabelle distribution and by [11].
The formal development accompanying this paper will be made available on the
web4 , which should also be consulted for proofs.
Before starting with the formal development, let us give an example of the
kind of program (see Fig. 1) that we would like to write. Assume a knowledge
base with objects of class A and B, and a relation r. The node n is initially
connected to at least 3 objects of class A, and all objects it is connected to are
of class B. Because the number of connections to A is too large, we execute a
loop that selects an A-object (let’s call it a) that n is connected to, and delete
the r-connection between n and a. To compensate, we select an object b of class
B and connect n to b. We stop as soon as the number of A-connections of n has
reached 2, which is one of the post-conditions we can ascertain.
2 Logic
Our logic is a three-tier framework, the first level being DL concepts (“TBox”),
the second level facts (“ABox”, instances of concepts), the third level formulas
(Boolean combinations of facts and a simple form of quantification).
Concepts: We concentrate on a DL featuring concepts with simple roles and
number restrictions, similar to ALCN [2]. For c being the type of concept names
4
http://www.irit.fr/~Martin.Strecker/Publications/dl_transfo2013.html
Logical Foundations for Reasoning about Transformations ... 523
vars n, a, b;
/* Pre: n : (≥ 3 r A) u (∀ r B) */
while ( n : (> 2 r A) ) do {
/* Inv: n : (≥ 2 r A) u (∀ r B) */
select a sth a : A ∧ (n r a);
delete(n r a);
select b sth b : B ;
add(n r b)
}
/* Post: n : (= 2 r A) u (∀ r B) */
Fig. 1. An example program
and r the type of role names, the data type C of concepts can be defined induc-
tively by:
C ::= c (atomic concept)
| ¬C (negation)
| C u C (conjunction)
| C t C (disjunction)
| (≥ n r C) (at least)
| (< n r C) (no more than)
| C[r := RE] (explicit substitution)
Adding a universal concept > and an empty concept ⊥ would not add expres-
sivity, as they are equivalent to (≥ 0 r C) respectively (< 0 r C) for arbitrary r
and C, and we will use them as shortcuts. We also write (∃ r C) for (≥ 1 r C)
and (∀ r C) for (< 1 r (¬C)).
The last constructor, explicit substitution [1], is a particularity of our frame-
work, required for a lazy elimination of substitutions that replace, in a concept
C, a role name r by a role expression RE. If i is the set of individual variable
names, the type RE is defined by
RE ::= r (atomic role)
| r − (i, i) (deletion of relation instance)
| r + (i, i) (insertion of relation instance)
Please note that concepts implicitly depend on the types c, r and i, which we
assume mutually disjoint. A substitution can therefore never affect an individual
variable.
A set-theoretic semantics is provided by a domain ∆ an interpretation func-
tion I mapping c to a set of individuals (subsets of ∆), r to a binary relation of
individuals (subsets of ∆ × ∆), and i to individual elements of ∆.
For interpretation of concepts C, negation is inductively interpreted as com-
plement, concept conjunction as intersection and disjunction as union. I(≥
524 M. Chaabani, R. Echahed and M. Strecker
n r C) = {x | card{y | (x, y) ∈ I(r) ∧ y ∈ I(C)} ≥ n}, and analogously
for I(< n r C). Here, card is the cardinality of finite sets (and 0 otherwise).
For interpretation of role expressions RE, we define I(r − (i1 , i2 )) = I(r) −
{(I(i1 ), I(i2 ))}, and I(r + (i1 , i2 )) = I(r) ∪ {(I(i1 ), I(i2 ))}.
Interpretation update I [r:=rl] modifies the interpretation I at relation name
r to relation rl, thus I [r:=rl] (r) = rl and I [r:=rl] (r0 ) = I(r0 ) for r0 6= r. With
this, we can define the semantics of explicit substitution by I(C[r := RE]) =
I [r:=I(RE)] (C).
Facts: Facts make assertions about an instance being an element of a concept,
and about being in a relation. In DL parlance, facts are elements of an ABox.
The type of facts is defined as follows:
f act ::= i : C (instance of concept)
| iri (instance of role)
| i (¬r) i (instance of role complement)
| i = i (equality of instances)
| i 6= i (inequality of instances)
The interpretation of a fact is a truth value, defined by:
– I(i : C) = (I(i) ∈ I(C))
– I(i1 r i2 ) = (I(i1 ), I(i2 )) ∈ I(r) and I(i1 (¬r) i2 ) = (I(i1 ), I(i2 )) ∈
/ I(r)
– I(i1 = i2 ) = (I(i1 ) = I(i2 )) and I(i1 6= i2 ) = (I(i1 ) 6= I(i2 ))
Please note that since concepts are closed by complement, facts are closed by
negation (the negation of a fact is again representable as a fact), and this is the
main motivation for introducing the constructors “instance of role complement”
and “inequality of instances”.
Formulas: A formula is a Boolean combination of facts. We also allow quantifi-
cation over individuals i (but not over relations or concepts), and, again, have a
constructor for explicit substitution.
f orm ::= ⊥
| f act
| ¬f orm
| f orm ∧ f orm | f orm ∨ f orm
| ∀i.f orm | ∃i.f orm
| f orm[r := RE]
The extension of interpretations from facts to formulas is standard; the inter-
pretation of substitution in formulas is in entire analogy to concepts. As usual,
a formula that is true under all interpretations is called valid.
When calculating weakest preconditions (in Sect. 4), we obtain formulas
which essentially contain no existential quantifiers; we keep them as constructor
because they can occur as intermediate result of computations. We say that a
formula is essentially universally quantified if ∀ only occurs below an even and ∃
only below an odd number of negations. For example, ¬(∃x. x : C ∧ ¬(∀y. y : D))
is essentially universally quantified.
Logical Foundations for Reasoning about Transformations ... 525
Implication f1 −→ f2 is the abbreviation for ¬f1 ∨ f2 , and ite(c, t, e) the
abbreviation for (c −→ t) ∧ (¬c −→ e), not to be confused with the if-then-else
statement presented in Sect. 3.
3 Programming Language
The programming language is an imperative language manipulating relational
structures. Its distinctive features are conditions (in conditional statements and
loops) that are restricted DL formulas, in the sense of Sect. 2. It has a non-
deterministic assignment statement allowing to select an element satisfying a
fact. Traditional types (numbers, inductive types) are not provided.
In this paper, we only consider a core language with traditional control flow
constructs, but without procedures. Also, it is only possible to modify a relational
structure, but not to “create objects” (with a sort of new statement) or to
“deallocate” them. These constructs are left for further investigation.
3.1 Syntax
The type of statements is defined by:
stmt ::= Skip (empty statement)
| select i sth f orm (assignment)
| delrel(i r i) (delete arc in relation)
| insrel(i r i) (insert arc in relation)
| stmt ; stmt (sequence)
| if f orm then stmt else stmt
| while f orm do stmt
3.2 Semantics
The semantics is a big-step semantics with rules of the form (st, σ) ⇒ σ 0 ex-
pressing that executing statement st in state σ produces a new state σ 0 .
The rules of the semantics are given in the Fig. 2. Beware that we overload
logical symbols such as ∃, ∧ and ¬ for use in the meta-syntax and as constructors
of f orm.
The state space σ is in fact identical to an interpretation function I as intro-
duced in Sect. 2, and it is only in keeping with traditional notation in semantics
that we use the symbol σ. We may therefore write σ(b) to evaluate the condition
b (a formula) in state σ.
Most of the rules are standard, apart from the fact that we do not use expres-
sions, but formulas as conditions. The auxiliary function delete edge modifies the
state σ by removing an r-edge between the elements represented by v1 and v2 .
With the update function for interpretations introduced in Sect. 2, one defines
delete edge v1 r v2 σ = σ [r:=σ(r)−{(σ(v1 ),σ(v2 ))}]
526 M. Chaabani, R. Echahed and M. Strecker
(c1 , σ) ⇒ σ 00 (c2 , σ 00 ) ⇒ σ 0
(Skip) (Seq)
(Skip, σ) ⇒ σ (c1 ;c2 , σ) ⇒ σ 0
σ 0 = delete edge v1 r v2 σ σ 0 = generate edge v1 r v2 σ
(EDel) (EGen)
(delrel(v1 r v2 ), σ) ⇒ σ 0 (insrel(v1 r v2 ), σ) ⇒ σ 0
∃vi.(σ 0 = σ [v:=vi] ∧ σ 0 (b))
(SelAssT )
(select v sth b, σ) ⇒ σ 0
σ(b) (c1 , σ) ⇒ σ 0 ¬σ(b) (c2 , σ) ⇒ σ 0
0
(If T ) (If F )
(if b then c1 else c2 , σ) ⇒ σ (if b then c1 else c2 , σ) ⇒ σ 0
σ(b) (c, σ) ⇒ σ 00 (while b do c, σ 00 ) ⇒ σ 0 ¬σ(b)
(W T ) (W F )
(while b do c, σ) ⇒ σ 0 (while b do c, σ) ⇒ σ
Fig. 2. Big-step semantics rules
and similarly
generate edge v1 r v2 σ = σ [r:=σ(r)∪{(σ(v1 ),σ(v2 ))}]
The statement select v sth F (v) selects an element vi that satisfies formula
F , and assigns it to v. For example, select a sth a : A∧(a r b) selects an element
a instance of concept A and being r-related with a given element b.
select is a generalization of a traditional assignment statement. There may
be several instances that satisfy F , and the expressiveness of the logic might
not suffice to distinguish them. In this case, any such element is selected, non-
deterministically. Let us spell out the precondition of (SelAssT ): Here, σ [v:=vi]
is an interpretation update for individuals, modifying σ at individual name v ∈ i
with an instance vi ∈ ∆, similar to the interpretation update for relations seen
before. We therefore pick an instance vi, check whether the formula b would be
satisfied under this choice, and if it is the case, keep this assignment.
In case no satisfying instance exists, the semantics blocks, i.e. the given state
does not have a successor state, which can be considered as an error situation.
Some alternatives to this design choice can be envisaged: We might treat a
select v sth F (v) with unsatisfiable F as equivalent to a Skip. This would
give us a choice of two rules, one in which the precondition of rule (SelAssT ) is
satisfied, and one in which it is not. As will be seen in Sect. 4, this would introduce
essentially existentially quantified variables in our formulas when computing
Logical Foundations for Reasoning about Transformations ... 527
weakest preconditions and lead us out of the fragment that we can deal with in
our decision procedure. Alternatively, we could apply an extended type check
verifying that select-predicates are always satisfiable, and thus ensure that type-
correct programs do not block. This is the alternative we prefer; details still have
to be worked out.
4 Weakest Preconditions
We compute weakest preconditions wp and verification conditions vc. Both take
a statement and a DL formula as argument and produce a DL formula. For this
purpose, while loops have to be annotated with loop invariants, and the while
constructor becomes: while {f orm} f orm do stmt. Here, the first formula (in
braces) is the invariant, the second formula the termination condition. The two
functions are defined by primitive recursion over statements, see Fig. 3.
wp(Skip, Q) = Q
wp(delrel(v1 r v2), Q) = Q[r := r − (v1 , v2 )]
wp(insrel(v1 r v2), Q) = Q[r := r + (v1 , v2 )]
wp(select v sth b, Q) = ∀v.(b −→ Q)
wp(c1 ; c2 , Q) = wp(c1 , wp(c2 , Q))
wp(if b then c1 else c2 , Q) = ite(b, wp(c1 , Q), wp(c2 , Q))
wp(while{iv} b do c, Q) = iv
vc(Skip, Q) = >
vc(delrel(v1 r v2), Q) = >
vc(insrel(v1 r v2), Q) = >
vc(select v sth b, Q) = >
vc(c1 ; c2 , Q) = vc(c1 , wp(c2 , Q)) ∧ vc(c2 , Q)
vc(if b then c1 else c2 , Q) = vc(c1 , Q) ∧ vc(c2 , Q)
vc(while{iv} b do c, Q) = (iv ∧ ¬b −→ Q) ∧ (iv ∧ b −→ wp(c, iv)) ∧ vc(c, iv)
Fig. 3. Weakest preconditions and verification conditions
Without going further into program semantics issues, let us only state the fol-
lowing soundness result that relates the operational semantics and the functions
wp and vc:
Theorem 1 (Soundness). If vc(c, Q) is valid and (c, σ) ⇒ σ 0 , then σ(wp(c, Q))
implies σ 0 (Q).
What is more relevant for our purposes is the structure of the formulas gener-
ated by wp and vc, because it has an impact on the decision procedure. Besides
the notion of essentially universally quantified introduced in Sect. 2, we need
528 M. Chaabani, R. Echahed and M. Strecker
the notion of quantifier-free formula: A formula not containing a quantifier. In
extension, we say that a statement is quantifier-free if all of its formulas are
quantifier-free.
By induction on c, one shows:
Lemma 1 (Universally quantified). Let Q be essentially universally quanti-
fied and c be a quantifier-free statement. Then wp(c, Q) and vc(c, Q) are essen-
tially universally quantified.
5 Decision Procedure
5.1 Overview
We present a decision procedure for verifying the validity of essentially univer-
sally quantified formulas. As seen in Lemma 1, this is the format of formulas
extracted by wp and vc, and as motivated by the soundness result (Theorem 1),
validity of verification conditions is a precondition for ensuring that a program
executes according to its specification.
Given an essentially universally quantified formula e, the rough lines of the
procedure for determining that e is valid are spelled out in the following.
Getting rid of quantifiers:
1. Convert e to an equivalent prenex normal form p, which will consist of a
prefix of universal quantifiers, and a quantifier-free body: ∀x1 . . . xn .b
2. p is valid iff its universal closure ucl(p) (universal abstraction over all free
variables of p) is.
3. Show the validity of ucl(p) by showing the unsatisfiability of ¬ucl(p).
4. ¬ucl(p) has the form ¬∀v1 . . . vk , x1 . . . xn .b. Pull negation inside the univer-
sal quantifier prefix, remove the resulting existential quantifier prefix, and
show unsatisfiability of ¬b with the aid of an extended tableau method.
Computation of prenex normal forms is standard. Care has to be taken to
avoid capture of free variables, by renaming bound variables. Free variables are
defined as usual; the free variables of a substitution f [r := r − (v1 , v2 )] are those
of f and in addition v1 and v2 (similarly for edge insertion). We illustrate the
problem with the following program fragment prg:
select a sth a : A ;
select b sth b r a ;
select a sth a r b
For a given post-condition Q, we obtain
wp(prg, Q) = ∀a.a : A −→ ∀b.(b r a) −→ ∀a.(a r b) −→ Q
whose prenex normal form ∀a1 , b, a2 . (a1 : A −→ (b r a1 ) −→ (a2 r b) −→ Q)
contains more logical variables than prg contains program variables.
Logical Foundations for Reasoning about Transformations ... 529
Extended tableau method – prerequisites: The tableau method takes a quantifier-
free formula f and proves its unsatisfiability or displays a model. We aim at
reusing existing tableau methods (such as [3]) as much as possible. The difficulty
consists in getting rid of the substitution constructor.
Substitution is compatible with the constructors of formulas:
Lemma 2 (Substitution in formulas).
⊥[r := re] = ⊥
(¬f )[r := re] = (¬f [r := re])
(f1 ∧ f2 )[r := re] = (f1 [r := re] ∧ f2 [r := re])
(f1 ∨ f2 )[r := re] = (f1 [r := re] ∨ f2 [r := re])
The case of formulas which are facts, missing in Lemma 2, will be dealt with
separately. This is due to the fact that substitution is not compatible with con-
cepts, as will be seen in Sect. 5.2: For a given concept C, there is not necessarily
a concept C 0 = C[r := re]. However, substitutions can be eliminated from facts,
by the equations given in Sect. 5.2.
We will refer to the equations in Lemma 2 and those in Sect. 5.2 as substitu-
tion elimination rules. We say that a substitution in a formula is visible if one of
these rules is applicable; and that it is hidden if none of these rules is applicable.
For example, the substitution in (x : (C1 u C2 ))[r := re] is visible; it is hidden
in (x : (C1 [r := re] u C2 [r := re])) and only becomes visible after application of
an appropriate tableau rule, for example of the system ALCN .
To describe our procedure, we introduce the following terminology: An ABox
is a finite set of facts (interpreted as the conjunction of its facts), and a tableau
a finite set of ABoxes (interpreted as a disjunction of its ABoxes). We need the
following functions:
– push subst takes a formula and applies substitution elimination rules as far
as possible;
– f orm to tab converts to disjunctive normal form and then performs the ob-
vious translation to a tableau;
– tab to f orm takes a tableau and constructs the corresponding formula.
Extended tableau method – procedure: Our method is parameterized by the fol-
lowing interface of an implementation of your favorite tableau calculus:
– a transition system T =⇒ T 0 , defining a one-step transformation of a tableau
T to a tableau T 0 .
– a function sat which checks, for tableaux T that are irreducible wrt. =⇒,
whether T is satisfiable.
From this, we construct a restricted relation T =⇒r T 0 , which is the same
as =⇒ provided that T does not contain visible substitutions:
T =⇒ T 0 no visible subst in T
T =⇒r T 0
530 M. Chaabani, R. Echahed and M. Strecker
We also define a relation =⇒s that pushes substitutions until they become
hidden:
T contains visible subst T 0 = f orm to tab(push subst(tab to f orm(T )))
T =⇒s T 0
From these, we define the relation =⇒sr = (=⇒r ∪ =⇒s ).
The extended tableau algorithm takes a formula f and computes a Tf such
that f orm to tab(f )(=⇒sr )∗ Tf . The result of the algorithm is sat(Tf ).
The following lemmas show that =⇒sr is a correct and complete algorithm
for deciding the decidability of formulas with substitution provided =⇒ is for
substitution-free formulas.
Lemma 3 (Termination). =⇒sr is well-founded provided =⇒ is.
To show termination of the extended algorithm, define
– the substitution size of a formula or fact as the sum of the term sizes below
its substitutions.
– the substitution size of a tableau as the multiset of the substitution sizes of
its facts.
Note that application of =⇒s leads to a reduction of the substitution size. For
a well-founded measure m of =⇒, construct a well-founded measure of =⇒sr as
the lexicographic order of the substitution size and m.
Lemma 4 (Confluence). =⇒sr is confluent provided =⇒ is.
=⇒sr has no other critical pairs than =⇒.
Lemma 5 (Satisfiability). =⇒sr preserves satisfiability provided =⇒ does.
The three auxiliary functions used for defining =⇒s do.
5.2 Elimination of Substitutions
We now show how substitutions can be pushed into facts.
The constructors equality and inequality are easiest to handle:
– (x = y)[r := re] reduces to (x = y)
– (x 6= y)[r := re] reduces to (x 6= y)
For positive resp. negative instances of roles, we have:
– (x r y)[r := r − (v1 , v2 )] reduces to (¬((x = v1 ) ∧ (y = v2 ))) ∧ (x r y)
– (x (¬r) y)[r := r − (v1 , v2 )] reduces to ((x = v1 ) ∧ (y = v2 )) ∨ (x (¬r) y)
– (x r y)[r := r + (v1 , v2 )] reduces to ((x = v1 ) ∧ (y = v2 )) ∨ (x r y)
– (x (¬r) y)[r := r + (v1 , v2 )] reduces to (¬((x = v1 ) ∧ (y = v2 ))) ∧ (x (¬r) y)
Logical Foundations for Reasoning about Transformations ... 531
whereas substitutions (x r y)[r0 := re] and (x (¬r) y)[r0 := re] for r 6= r0 are
the identity.
For facts of the form x : C, where C is a concept, we have the cases:
– (x : ¬C)[r := re] reduces to x : (¬C[r := re])
– (x : C1 ∧ C2 )[r := re] reduces to x : (C1 [r := re] ∧ C2 [r := re])
– (x : C1 ∨ C2 )[r := re] reduces to x : (C1 [r := re] ∨ C2 [r := re])
– (x : (≥ n r C))[r0 := re], for r0 6= r, reduces to x : (≥ n r C[r0 := re]), and
similarly when replacing ≥ by <
– (x : (≥ n r C))[r := r − (v1 , v2 )] reduces to
ite ((x = v1 ) ∧ (v2 : (C[r := r − (v1 , v2 )])) ∧ (v1 r v2 ),
(x : (≥ (n + 1) r (C[r := r − (v1 , v2 )]))),
(x : (≥ n r (C[r := r − (v1 , v2 )]))))
and similarly when replacing ≥ by <
– (x : (≥ (n + 1) r C))[r := r + (v1 , v2 )] reduces to
ite ((x = v1 ) ∧ (v2 : (C[r := r + (v1 , v2 )])) ∧ (v1 (¬r) v2 ),
(x : (≥ n r (C[r := r + (v1 , v2 )]))),
(x : (≥ (n + 1) r (C[r := r + (v1 , v2 )]))))
and similarly when replacing ≥ by <
– (x : (≥ 0 r C))[r := r + (v1 , v2 )] reduces to >
– (x : (< 0 r C))[r := r + (v1 , v2 )] reduces to ⊥
– Pathological case (x : C[sbst1 ])[sbst2 ]: lift inner substitution to (x : C)[sbst1 ][sbst2 ],
then apply the above.
6 Conclusions
This paper proposes a language for rewriting knowledge bases, and methods for
reasoning about the correctness of these programs, by means of a Hoare-style
calculus. DL formulas are directly integrated into the statements of the pro-
gramming language. The verification conditions extracted from these programs
has been shown to be decidable, by a modular extension of existing tableau
algorithms.
The work described here is still preliminary, in several respects, and the
following points indicate directions for future investigations:
– We are in the process of coding the theory in the Isabelle proof assistant.
Some parts of the proofs of Sect. 4 and most of Sect. 5.1 still has to be done.
The purpose is to obtain a framework that will allow us to experiment more
easily with variations of the logic.
– We have currently focused on the logic ALCN . It is interesting to consider
both less expressive logics (which offer more space for optimizations) and
more expressive logics (to explore decidability questions). The process de-
scribed in Sect. 5.1 is rather generic, but it remains to be seen whether more
expressive DLs, featuring more complex role expressions, can be accommo-
dated.
532 M. Chaabani, R. Echahed and M. Strecker
– In any case, the proof procedure sketched in Sect. 5 is rather of a theoret-
ical than a practical value; an efficient implementation should not convert
between formulas and tableaux as indiscriminately as suggested there, but
apply propagation of substitutions locally.
– In a similar vein, it would be interesting to implement a transformation
engine on the basis of the language described here, also with the purpose of
evaluating the practical expressiveness of the language on larger examples.
References
1. Abadi, M., Cardelli, L., Curien, P.L., Lévy, J.J.: Explicit substitutions. Journal of
Functional Programming 1(4), 375–416 (October 1991)
2. Baader, F., Sattler, U.: Expressive number restrictions in description logics. Journal
of Logic and Computation 9(3), 319–350 (1999)
3. Baader, F., Sattler, U.: Tableau algorithms for description logics. In: Dyckhoff, R.
(ed.) Automated Reasoning with Analytic Tableaux and Related Methods, Lecture
Notes in Computer Science, vol. 1847, pp. 1–18. Springer Berlin / Heidelberg (2000)
4. Balbiani, P., Echahed, R., Herzig, A.: A dynamic logic for termgraph rewriting. In:
5th International Conference on Graph Transformations (ICGT). Lecture Notes in
Computer Science, vol. 6372, pp. 59–74. Springer (2010)
5. Chaabani, M., Mezghiche, M., Strecker, M.: Vérification d’une méthode de preuve
pour la logique de description ALC. In: Ait-Ameur, Y. (ed.) Proc. 10ème Journées
Approches Formelles dans l’Assistance au Développement de Logiciels (AFADL).
pp. 149–163 (Jun 2010)
6. Courcelle, B., Engelfriet, J.: Graph structure and monadic second-order logic, a
language theoretic approach. Cambridge University Press (2011)
7. Immerman, N., Rabinovich, A., Reps, T., Sagiv, M., Yorsh, G.: The bound-
ary between decidability and undecidability for transitive-closure logics. In:
Marcinkowski, J., Tarlecki, A. (eds.) Computer Science Logic, Lecture Notes in
Computer Science, vol. 3210, pp. 160–174. Springer Berlin / Heidelberg (2004)
8. Liu, H., Lutz, C., Milicic, M., Wolter, F.: Foundations of instance level updates in
expressive description logics. Artificial Intelligence 175(18), 2170–2197 (2011)
9. Møller, A., Schwartzbach, M.I.: The pointer assertion logic engine. In: PLDI. pp.
221–231 (2001)
10. Nipkow, T., Paulson, L., Wenzel, M.: Isabelle/HOL. A Proof Assistant for Higher-
Order Logic, Lecture Notes in Computer Science, vol. 2283. Springer Berlin /
Heidelberg (2002)
11. Schirmer, N.: Verification of Sequential Imperative Programs in Isabelle/HOL.
Ph.D. thesis, Technische Universität München (2006)
Program Algebras with Monotone Floyd-Hoare
Composition
Andrii Kryvolap1, Mykola Nikitchenko1 and Wolfgang Schreiner2
1
Taras Shevchenko National University of Kyiv, Kyiv, Ukraine
krivolapa@gmail.com, nikitchenko@unicyb.kiev.ua
2
Johannes Kepler University, Linz, Austria
Wolfgang.Schreiner@risc.jku.at
Abstract. In the paper special program algebras of partial predicates and func-
tions are described. Such algebras form a semantic component of a modified
Floyd-Hoare logic constructed on the base of a composition-nominative ap-
proach. According to this approach, Floyd-Hoare assertions are presented with
the help of a special composition called Floyd-Hoare composition. Monotonic-
ity and continuity of this composition are proved. The language of the modified
Floyd-Hoare logic is described. Further, the inference rules for such logic are
studied, their soundness conditions are specified. The logic constructed can be
used for program verification.
Keywords. Program algebra, program logic, composition-nominative approach,
partial predicate, soundness
Key terms. FormalMethod, VerificationProcess
1 Introduction
Program logics are the main formalisms used for proving assertions about program
properties. A well-known Floyd-Hoare logic [1, 2] is an example of such logics. Se-
mantically, this logic is defined for a case of total predicates and functions though
programs can be partial. In this case assertions can be presented with the help of a
special composition over total predicates and functions called Floyd-Hoare composi-
tion (FH-composition). However, a straightforward extension of classical Floyd-
Hoare logic for partial predicates and functions meets some difficulties. The first one
is that the classical FH-composition will not be monotone. Monotonicity means that
the result of the mapping evaluation remains the same on extended data, if it was
evaluated on the initial data. This important property grants the possibility to reason
about the correctness of the program based on the correctness of its approximations.
534 A. Kryvolap, M. Nikitchenko and W. Schreiner
That is why the need of a modified definition of the classical Floyd-Hoare logic for
the case of partial mappings arises. Here we will consider only mappings (predicates,
ordinary functions, and program functions) defined over flat nominative data (nomi-
native sets). Such data are treated as collections of named values. Mappings over such
data are called quasiary mappings [3]. The obtained program algebras are called qua-
siary program algebras. They form a semantic component of quasiary Floyd-Hoare
logics.
The syntactic component of such logics is presented by their languages and sys-
tems of inference rules. We study the possibility to use classical rules for modified
logics with a monotone Floyd-Hoare composition. Systems of such inference rules
should be sound and complete to be of a practical use. This could be achieved by
adding proper restrictions to the inference rules of the classical Floyd-Hoare logic that
fail to be correct. It should be also shown that by weakening additional restrictions we
obtain a system of the inference rules that is not sound. This will prove that restric-
tions are necessary.
The rest of the paper is structured as follows. In Section 2 we describe program al-
gebras of quasiary predicates and functions on different levels of abstraction, define a
modified Floyd-Hoare composition and specify the syntax for the modified logic. In
Section 3 we prove the main properties of this composition. In Section 4 we study the
soundness of the system of inference rules for the introduced program algebras. Fi-
nally, we formulate conclusions in Section 5.
2 Quasiary Program Algebras
To modify the classical Floyd-Hoare logic for partial quasiary mappings, we will use
semantic-syntactic scheme [3-5]. This means that we will first define the semantics in
the form of classes of quasiary program algebras. Then the language of the logic will
be defined as well as the interpretation mappings.
p
To emphasize a mapping’s partiality/totality we write the sign for partial
t
mappings and the sign
for total mappings. Given an arbitrary partial mapping
p
: D D , d D, S D, S D we write:
– (d) to denote that is defined on d;
– (d)= d to denote that is defined on d with a value d ;
– (d) to denote that is undefined on d;
– μ[ S ] {μ(d ) | μ(d ) , d S } to denote the image of S under ;
– μ 1[ S ' ] {d | μ(d ) , μ(d ) S ' } to denote the preimage (inverse image) of S
under .
2.1 Classes of quasiary mappings
Let V be a set of names (variables). Let A be a set of basic values. Given V and A, the
class VA of nominative sets is defined as the class of all partial mappings from V to A,
Program Algebras with Monotone Floyd-Hoare Composition 535
p
thus, VA=V A. Informally speaking, nominative sets represent states of vari-
ables.
Though nominative sets are defined as mappings, we follow mathematical tradi-
tions and also use a set-like notation for these objects. In particular, the notation
d = [vi ai | iI] describes a nominative set d where vi ai n d means that d(vi) is
defined and its value is ai (d(vi)=ai). The main operation for nominative sets is the
t
binary total overriding operation : VA× VA VA defined by the formula
d1d 2 [v a | v a n d 2 (v a n d1 a(v a n d 2 ))] . Intuitively,
given d1 and d2 this operation yields a new nominative set which consists of named
pairs of d2 and those pairs of d1 whose names do not occur in d2.
p
Let Bool {F ,T } be the set of Boolean values. Let PrV, A=VA Bool be the
set of all partial predicates over VA. Such predicates are called partial quasiary predi-
p
cates. Let FnV, A=VA A be the set of all partial functions from VA to A. Such
functions are called partial quasiary ordinary functions. Here ‘ordinary’ means that
p
the range of such functions is the set of basic values A. Let FPrgV, A=VA VA be
the set of all partial functions from VA to VA. Such functions are called bi-quasiary
functions.
Quasiary predicates represent conditions which occur in programs, quasiary ordi-
nary functions represent the semantics of program expressions, and biquasiary func-
tions represent program semantics.
The terms ‘partial’ and ‘ordinary’ are usually omitted. In a general term, elements
from PrV, A, FnV, A, and FPrgV, A are called quasiary mappings.
2.2 Hierarchy of program algebras and logics
Based on algebras with three carriers (PrV, A, FnV, A, and FPrgV, A) we can
define logics of three types:
– Pure quasiary predicate logics based on algebras with one sort: PrV,А
– Quasiary predicate-function logics based on algebras with two sorts: Pr V,А and
FnV,А
– Quasiary program logics based on algebras with three sorts: PrV,А, FnV,А, and
FPrgV,А
For logics of pure quasiary predicates we identify renominative, quantifier, and
quantifier-equational levels.
Renominative logics [3] are the most abstract among above-mentioned logics. The
main new compositions for these logics are the compositions of renomination (renam-
v ,...,v
ing) of the form R x1 ,..., xn : PrV,А
t
PrV,А. Intuitively, given a quasiary predicate p
1 n
v ,...,v
and a nominative set d the value of R x1 ,..., xn (p)(d) is evaluated in the following way:
1 n
first, a new nominative set d is constructed from d by changing the values of the
536 A. Kryvolap, M. Nikitchenko and W. Schreiner
names v1,...,vn in d to the values of the names x1,..., xn respectively; then the predicate
p is applied to d . The obtained value (if it was evaluated) will be the result of
v ,...,v
R x1 ,..., xn (p)(d). For this composition we will also use a simplified notation Rxv . The
1 n
basic compositions of renominative logics are , , and R vx . Note, that renomination
(primarily in syntactical aspects) is widely used in classical logic, lambda-calculus, and
specification languages like Z-notation, B, TLA, RAISE, ASM, etc.
At the quantifier level, all basic values can be used to construct different nomina-
tive sets to which quasiary predicates can be applied. This allows one to introduce the
compositions of quantification of the form x in style of Kleene’s strong quantifiers.
The basic compositions of logics of the quantifier level are , , R vx , and x.
At the quantifier-equational level, new possibilities arise for equating and differen-
tiating values with special 0-ary compositions of the form =xy called equality predi-
cates. Basic compositions of logics of the quantifier-equational level are , , Rxv ,
x, and =ху .
All specified logics (renominative, quantifier, and quantifier-equational) are based
on algebras that have only one sort: a class of quasiary predicates.
For quasiary predicate-function logics we identify the function level and the func-
tion-equational level.
At the function level, we have extended capabilities for the formation of new ar-
guments of functions and predicates. In this case it is possible to introduce the super-
position compositions S Fv and S Pv (see [4, 5]), which formalize substitution of func-
tions into function and predicate respectively. Also special null-ary denomination
parametric compositions (functions) 'x are introduced. The introduction of such func-
tions allows one to model renomination compositions with the help of superpositions.
The basic compositions of logics of the function level are , , S Fv , S Pv , x, and 'x.
At the function-equational level, a special equality composition = can be intro-
duced additionally. The basic compositions of logics of the function-equational level
are , , S Fv , S Pv , x, 'x, and = . At this level different classes of first-order logics
can be presented.
This means that two-sorted algebras (with sets of predicates and functions as sorts
and above-mentioned compositions as operations) form a semantic base for first-order
CNL.
The level of program logics is quite rich. Investigation of such logics is a special
challenge; here we will study semantic properties of a modified Floyd-Hoare logic.
To define such logics we should first define program algebras with program composi-
tions as their operations. Such compositions correspond to the main structures of pro-
grams. In the simplest case they are:
– The parametric assignment composition AS x : FnV , A FPrg V , A
– The composition of sequential execution : FPrg V , A FPrg V , A FPrg V , A
– The conditional composition IF : PrV , A FPrg V , A FPrg V , A FPrg V , A
– The cyclic composition (loop) WH : Pr V , A FPrg V , A FPrg V , A
Program Algebras with Monotone Floyd-Hoare Composition 537
Additionally we need compositions that describe properties of the programs. The
Floyd-Hoare composition FH : PrV , A FPrg V , A Pr V , A PrV , A is the most important
of them. Its formal definition will be given in the next subsection.
2.3 Formal definition of a Floyd-Hoare composition
The required definition stems from the treatment of Floyd-Hoare assertions with total
predicates (see, for example, [6]). Namely, an assertion {p}f{q} is said to be valid if
and only if
for all d from VA if p(d) =T, f(d)= d for some d then q(d ) =T (1)
Note, that we do not make a distinction between a formula and its interpretation.
Thus, we treat, say, p as a formula in the assertion {p}f{q} and as a predicate of the
program algebra.
The definition (1) permits to treat {p}f{q} as a predicate because this is a pointwise
definition. Rewriting this definition for different cases we get the following matrices
(table 1) specifying the logical values of {p}f{q} for an arbitrarily d:
Table 1. Logical values of {p}f{q} for total predicates.
a) f(d) is defined b) f(d) is undefined
p(d) \ q(f(d)) F T p(d) {p}f{q}(d)
F T T F T
T F T T T
Our aim is to extend the notion of assertion validity for partial predicates. But first
we should admit that the presented definition will not be monotone under predicate
extension. Indeed, consider informally the following assertion:
{T} while T do skip {F}.
This Floyd-Hoare triple will be true on all data, because the infinite loop is
undefined on all data, and thus on all data the condition of validity for this assertion is
satisfied. Now consider a triple {T} skip {F} that is false on all data. However, the
mapping ‘skip’ is an extension of ‘while T do skip’. Thus, monotonicity fails for a
case when p(d)=T and f(d) is undefined. So, the value for this case should be changed.
To define a monotone interpretation of Floyd-Hoare triple for partial predicates we
should change the question marks in Table 2 to Boolean values.
538 A. Kryvolap, M. Nikitchenko and W. Schreiner
Table 2. Logical values of {p}f{q} for partial predicates, where the question marks represent
values that should be changed to proper Boolean values.
a) f(d) is defined b) f(d) is undefined
p(d) \ q(f(d)) F T Undefined p(d) {p}f{q}(d)
F T T ? F T
T F T ? T ?
undefined ? ? ? undefined ?
To define such interpretation we adopt the following requirements:
– Monotonicity of a composition on all its arguments
– Maximal definiteness of the obtained predicates (we call this requirements as
Kleene’s principle)
We use techniques for non-deterministic semantics described in [7]. We treat the
case when a predicate is ‘undefined’ as non-deterministic values T and F. Thus, we
can use matrices from table 1 to evaluate a set of Boolean values for every case. The
obtained results are presented in Table 3.
Table 3. Logical values of {p}f{q} for partial predicates presented as sets of Boolean values.
a) f(d) is defined b) f(d) is undefined
p(d) \ q(f(d)) {F} {T} {F,T} p(d) {p}f{q}(d)
{F} {T} {T} {T} {F} {T}
{T} {F} {T} {F,T} {T} {F,T}
{F,T} {F,T} {T} {F,T} {F,T} {F,T}
Now, replacing non-deterministic results {F, T} on undefined we get the final results
(table 4).
Table 4. Logical values of {p}f{q} for partial predicates.
a) f(d) is defined b) f(d) is undefined
p(d)\q(f(d)) F T undefined p(d) {p}f{q}(d)
F T T T F T
T F T undefined T undefined
undefined undefined T undefined undefined undefined
The obtained matrices define an interpretation of {p}f{q} for partial predicates. As
was said earlier, we formalize such triples as a Floyd-Hoare composition
FH : PrV , A FPrg V , A Pr V , A PrV , A (p, q PrV,A, fFPrgV,А, d VA):
T , if q( f (d )) T or p(d ) F ,
FH(p,f, q)(d)= F , if p(d ) T and q( f (d )) F ,
undefined in other cases.
Program Algebras with Monotone Floyd-Hoare Composition 539
2.4 Formal definition of program algebra compositions
In the previous subsection the formal definition of FH-composition was presented. In
this subsection we give brief definitions of other compositions (see details in [3-5]).
Propositional compositions are defined as follows (p, q PrV,A, d VA):
T , if p(d ) T or q(d ) T , T , if p (d ) F ,
( p q)(d ) F , if p(d ) F and q(d ) F , (p)(d ) F , if p(d ) T ,
undefined in other cases. undefined if p(d ) .
Unary parametric composition of existential quantification x with the parameter
xV is defined by the following formula (p PrV,A, d VA):
T , if b A exists : p ( dx b) T ,
(x p )( d ) F , p ( dx a ) F for each a A,
undefined in other cases.
Here dx a is a shorter form for d[ x a] .
Parametric n-ary superpositions with x ( x1 ,..., xn ) as the parameter are defined
by the following formulas (f, g1,…, gn FnV,A, p PrV,A, d VA):
( S Fx ( f , g1 , , g n ))(d ) f (d [ x1 g1 (d ), , xn g n (d )]) ,
( S Px ( p, g1 , , g n ))( st ) f ( st[ x1 g1 ( st ), , xn g n ( st )]) .
Null-ary parametric denomination composition with the parameter xV is defined
by the following formula (d VA): 'x (d) = d(x).
Binary equality composition = is defined as follows (f, g FnV,A, d VA):
T , if f (d ) , g (d ) , and f (d ) g (d ),
(f=g) (d) F , if f (d ) , g (d ) , and f (d ) g (d ),
undefined in other cases.
Identical program composition idFPrgV,А is the most simple: id (d ) d (d VA).
Assignment composition is defined as follows (f FnV,A, d VA):
AS x ( f )(d ) d [ x f (d )] .
Sequential execution is introduced in the ordinary way (fs1, fs2FPrgV,А, d VA):
fs1 fs2 (d ) fs2 ( fs1 (d )) .
Note, that we define by commuting arguments of conventional functional compo-
sition: fs1 fs2 fs2 fs1 .
Conditional composition depends on the value of the first function which is the
condition itself (p PrV,A, fs1, fs2FPrgV,А, d VA):
fs1 (d ), if p(d ) T ,
IF ( p, fs1 , fs2 )(d ) fs2 (d ), if p(d ) F ,
undefined in other cases.
540 A. Kryvolap, M. Nikitchenko and W. Schreiner
Cycle is defined by the following formulas: WH ( p, fs)(d ) d n , where d0 d ,
d1 fs(d0 ) , …, d n fs(d n 1 ) , moreover p(d 0 ) T , p(d1 ) T , … ,
V,A V,А V
p(d n 1 ) T ,….. p(d n ) F (p Pr , fsFPrg , d A).
It means that we have defined the following quasiary program algebra:
QPA(V, A) = < PrV,A, FnV,A, FPrgV,A; , , S Fv , S Pv , x, x, =, id, AS x, , IF, WH, FH>.
This algebra is the main object of our investigation.
2.5 Formal definition of program algebra terms
Terms of the algebra QPA(V, A) defined over sets of predicate symbols Ps, function
symbols Fs, program symbols Prs, and variables V specify the syntax (the language)
of the logic. We now give inductive definitions for terms Tr ( Ps, Fs, Prs,V ) , formulas
Fr ( Ps, Fs, Prs,V ) , program texts Pt ( Ps, Fs, Prs,V ) , and Floyd-Hoare assertions
FHFr ( Ps, Fs, Prs,V ) .
First we will define terms:
– if f Fs then f Tr ( Ps, Fs, Prs,V )
– if v V then ' v Tr ( Ps, Fs, Prs,V )
– if f Fs , t1 ,, tn Tr ( Ps, Fs, Prs,V ) , and v1 , , vn V are distinct variables
then S Fv ( f , t1 , , tn ) Tr ( Ps, Fs, Prs,V )
Then we will define program texts:
– id Pt ( Ps, Fs, Prs,V )
– if p Prs then p Pt ( Ps, Fs, Prs, V )
– if v V and t Tr ( Ps, Fs, Prs,V ) then AS v (t ) Pt ( Ps, Fs, Prs,V )
– if p1 , p2 Pt ( Ps, Fs, Prs,V ) then p1 p2 Pt ( Ps, Fs, Prs,V )
– if p1 , p2 Pt ( Ps, Fs, Prs,V ) and b Fr ( Ps, Fs, Prs,V ) then
IF (b, p1 , p2 ) Pt ( Ps, Fs, Prs,V )
– if p Pt ( Ps, Fs, Prs,V ) and b Fr ( Ps, Fs, Prs,V ) then
WH (b, p) Pt ( Ps, Fs, Prs,V )
Finally, formulas and Floyd-Hoare triples are defined:
– if p Ps then p Fr ( Ps, Fs, Prs,V )
– if Fr ( Ps, Fs, Prs,V ) then Fr ( Ps, Fs, Prs,V )
– if t1 , t2 Tr ( Ps, Fs, Prs,V ) then t1 t2 Fr ( Ps, Fs, Prs,V )
– if Fr ( Ps, Fs, Prs,V ) and v V then v Fr ( Ps, Fs, Prs,V )
– if , Fr ( Ps, Fs, Prs,V ) then Fr ( Ps, Fs, Prs,V );
Program Algebras with Monotone Floyd-Hoare Composition 541
– if p Ps , t1 ,, tn Tr ( Ps, Fs, Prs,V ) , and v1 , , vn V are distinct variables
then S Pv ( p, t1 , , tn ) Fr ( Ps, Fs, Prs, V )
– if f Pt ( Ps, Fs, Prs,V ) and p, q Fr ( Ps, Fs, Prs,V ) then
{ p} f {q} FHFr ( Ps, Fs, Prs,V )
After syntax and semantics have been defined, we need to specify the interpretation
mappings, assuming that interpretation mappings for the predicate symbols
I Ps : Ps Pr V , A , functional symbols I Fs : Fs FnV , A , and program symbols
I Prs : Prs FPrg V , A are given. Let J Fr : Fr ( Fs, Ps, Prs, V ) Pr V , A denote an inter-
pretation mapping for formulas, J Tr : Tr ( Fs, Ps, Prs,V ) FnV , A denote an interpreta-
tion mapping for terms and J Pt : Pt ( Fs, Ps, Prs,V ) Prg V , A denote an interpretation
mapping for programs. They are all defined in a natural way, only the case with asser-
tion needs special consideration:
J FHFr ({ p} f {q}) FH ( J Fr ( p), J P t ( f ), J Fr (q )) .
An assertion is said to be valid (denoted | { p} f {q} ) if a corresponding predicate
is not refutable.
3 Monotonicity and Continuity of the Floyd-Hoare
Composition
In the previous section, a function-theoretic style of composition definitions was used.
To prove properties of the FH-composition, it is more convenient to use a set-
theoretic style of definition.
The following sets are called respectively truth, false, and undefiniteness domains
of the predicate p over D:
pT {d | p(d ) T } ,
p F {d | p(d ) F } ,
p {d | p(d ) } .
The following definitions introduce various images and preimages involved in
Floyd-Hoare composition:
q T , f f 1[qT ] ,
q F , f f 1[q F ] ,
q , f f 1[q ] ,
p T , f f [ pT ] ,
pF, f f [ pF ] ,
p , f f [ p ] .
Using these notations we can define FH-composition by describing the truth and
false domains of the predicate that is the value of the composition:
542 A. Kryvolap, M. Nikitchenko and W. Schreiner
FH ( p, f , q)T p F q T , f ,
FH ( p, f , q) F pT q F , f .
Validity of formulas (predicates) is considered as irrefutability, that is
| p p F . From this follows that
| FH ( p, f , q ) pT q F , f .
Let us give a formal definition of the monotone composition.
Composition C : ( FPrg V , A )n ( PrV , A ) k ( FnV , A )m PrV , A is called monotone if
the following condition holds for all arguments of C:
f1 g1 , , f n g n , p1 q1 , , pk qk , a1 b1 , , am bm
C ( f1 , , f n , p1 , , pk , a1 , , am ) C ( g1 , , g n , q1 , , qk , b1 , , bm ) .
Theorem 1. Floyd-Hoare composition is monotone on every argument.
Let us prove monotonicity on every argument separately, examining their truth and
false domains. For truth domain we have:
p1 p2 p1T p2T p1T q F , f p2T q F , f
FH ( p1 , f , q) F FH ( p2 , f , q) F .
Similar, for the false domain of the precondition we have:
p1 p2 p1F p2F p1F q T , f p2 F q T , f
FH ( p1 , f , q)T FH ( p2 , f , q)T .
Thus, p1 p2 FH ( p1 , f , q) FH ( p2 , f , q) .
In the case of truth domain of postcondition the proof is similar:
q1 q2 q1T q2T q1T , f q2T , f p F q1T , f p F q2 T , f
FH ( p, f , q1 )T FH ( p, f , q2 )T .
The same for the false domain of postcondition:
q1 q2 q1F q2F q1 F , f q2 F , f pT q1 F , f pT q2 F , f
FH ( p, f , q1 ) F FH ( p, f , q2 ) F .
Thus, q1 q2 FH ( p, f , q1 ) FH ( p, f , q2 ) .
Let us show the monotonicity of the truth domains for the FP-composition:
f1 f 2 q T , f1 q T , f2 p F q T , f1 p F q T , f2
FH ( p, f1 , q )T FH ( p, f 2 , q)T .
Similar, for the false domains:
f1 f 2 q F , f1 q F , f2 pT q F , f1 pT q F , f2
FH ( p, f1 , q ) F FH ( p, f 2 , q ) F .
Also f1 f 2 FH ( p, f1 , q) FH ( p, f 2 , q) .
Thus, it was shown that the composition is monotone on every component, what is
needed to be proved.
For the constructed composition even stronger result is true, it is continuous. To
show this, the following definitions are made and the notion of continuity is given
(see, for example, [6]).
Program Algebras with Monotone Floyd-Hoare Composition 543
An infinite set of indexed functions (predicates) { f 0 , f1 ,}, f i f i 1 , i is
called a chain of functions (predicates).
The supremum of the above-mentioned set of indexed functions (predicates) is
called limit of the chain of functions (predicates), denoted as f i .
i
The composition C : ( Prg ) ( Pr
V ,A n
) ( Fn
V,A m
) Pr
V ,A l V,A
is called continuous
on the first argument if for arbitrary chain { fi | i } the following property
holds: C ( f i , g 2 , , g n , p1 , pm , q1 , ql ) C ( f i , g 2 , , g n , p1 , pm , q1 , ql ) .
i i
Continuity on the other arguments is defined in a similar manner.
Theorem 2. Floyd-Hoare composition is continuous on every argument.
Though this result follows from the general consideration, we give here its direct
proof. Let us show the continuity on the first argument. In the case of other arguments
the proof will be similar.
Consider a chain of predicates { pi | i } . Since Floyd-Hoare composition is
monotone, {FH ( pi , f , q) | i } will also be a chain. We need to show
that FH ( pi , f , q ) FH ( pi , f , q ) .
i i
For the arbitrary data d , there are two different possibilities –
pi (d ) and pi (d ) . In the first case none of the elements of the chain in de-
i i
fined on d . Thus j , FH ( pi , f , q )( d ) FH ( p j , f , q )( d ) , therefore needed
i
equality is obvious. If the limit is defined on these data, an element of the chain that is
also defined on this data could be found. Otherwise the limit would have been unde-
fined on those data, what is guaranteed by the inclusion relation on the elements of
the chain. Let the limit be the element with index k. Then
FH ( pi , f , q)(d ) = FH ( pk , f , q)(d ) and
i
FH ( pk , f , q )(d ) FH ( pi , f , q )(d ) ,
i
since i k , pi (d ) pk (d ) from the definition of the chain.
The following equality is obtained: FH ( pi , f , q )( d ) FH ( pi , f , q )( d ) .
i i
Since the data was chosen arbitrary, we get FH ( pi , f , q) FH ( pi , f , q ) , what
i i
was needed to be proved.
The proof for the other arguments (a program and a postcondition) is similar. Thus,
it is proven that the monotone Floyd-Hoare composition is also continuous on every
argument.
544 A. Kryvolap, M. Nikitchenko and W. Schreiner
4 Soundness of Inference Rules System in Floyd-Hoare
Algebras
In this section we adopt the same convention as earlier that we do not distinguish
between syntactic and semantic notation for formulas. We also assume that the alge-
bra QPA(V, A) is fixed and interpretation mappings are also fixed.
Since a result of the Floyd-Hoare composition can be undefined on some data,
classical inference rules can be unsound. This informally means that with true pre-
conditions they could give false postconditions. This happens because predicates can
be partial and compositions are defined in a way that differs from the classical Floyd-
Hoare composition to be monotone. Let us examine the following system of inference
rules to find out what conditions are required for rules to be sound:
{S [ x ] ( p, f )}AS x ( f ){ p} – Ax_AS
{ p}id{ p} – Ax_ID
{ p} f {q},{q}g{r}
– Ax_SEQ
{ p} f g{r}
{b p} f {q},{b p}g{q}
– Ax_IF
{ p}IF (b, f , g ){q}
{b p} f { p}
– Ax_WH
{ p}WH (b, f ){b p}
{ p } f {q }
– Ax_CONS
{ p} f {q}
Note that we do not include additional conditions for the consequence rule, be-
cause in different classes of algebras we will have different conditions.
An assertion { p} f {q} is said to be derived if there exists its derivation tree with
rules of the type Ax_AS, Ax_ID on its leaves. Derivability is denoted as | { p} f {q} .
Let us show that for the rules Ax_SEQ, Ax_WH , and Ax_CONS without addi-
tional conditions we can give such an example of the application of the inference rule
that will have true preconditions and false postconditions.
Consider Ax_SEQ with violation of the condition pT , f qT .
If this condition fails then p, q, f , d : p (d ) T , q ( f (d )) ,| { p} f {q} . In this case
we will take such r and g that | {q}g{r} and g ( f (d )) , r ( g ( f (d ))) F . This is
possible if we define them in the following way:
T , x f (d ),
g id , r ( x)
F , x f (d ).
Then | { p} f g{r} does not hold, while p(d ) T and r ( f g (d )) F , what is
equal to {d } pT r F , f g .
Consider Ax_WH with violation of the condition (b p )T , f pT .
We will construct such b , f , and p that the following properties hold:
Program Algebras with Monotone Floyd-Hoare Composition 545
| {b p} f { p}, (b p )T , f pT ,
| { p}WH (b, f ){b p} .
Let d1 d 2 d3 . Then b , f , and p are defined in the following manner:
T , x d3 ,
b( x )
F , x d3 .
x, x d1 , d 2 ,
f ( x) d 2 , x d1 ,
d ,x d .
3 2
T , x d 2 , d 3 ,
p ( x ) , x d 2 ,
F, x d .
3
It is not hard to check that the above-mentioned properties are not satisfied:
d 2 (b p)T , f , d 2 pT ,
d1 pT , d3 WH (b, f )(d1 ) ,
d3 (b p) F d1 (b p) F ,WH (b, f ) ,
d1 (b p )T , d 2 p , d3 p F .
Thus, | {b p} f { p} because for other data p is true.
That proves that the additional condition is necessary because in other cases the
rule is not sound while used on such examples.
The case with the rule Ax_CONS is similar to the previous one with the conditions
pT pT , q F qF .
So, it was shown that additional conditions are not redundant. Let us show that if
additional conditions hold then the rules are sound.
Theorem 3. Inference rules are sound with additional conditions. In other words:
| {S [ x ] ( p, f )} AS x ( f ){ p} ,
| { p}id{ p} ,
| { p} f {q} | {q}g{r} pT , f qT | { p} f g{r} ,
| {b p} f {q} | {b p}g{q} | { p}IF (b, f , g ){q} ,
| {b p} f { p} (b p)T , f pT | { p} WH (b, f ){b p} ,
| { p } f {q} pT pT q F qF | { p} f {q} .
Let us prove this for each rule.
For | {S [ x ] ( p, f )} AS x ( f ){ p} to hold it is needed that the following condition
x
holds: FH ( S [ x ] ( p, f ), AS x ( f ), p) F ( S [ x ] ( p, f ))T p F , AS ( f ) .
Assume that it is false and the intersection is not empty. Let some data d belongs
to the intersection.
If d ( S [ x ] ( p, f ))T then p(d [ x f (d )]) T .
546 A. Kryvolap, M. Nikitchenko and W. Schreiner
x
Let d p F , AS ( f ) then p( AS x ( f )(d )) p(d [ x f (d )]) F , what is impossi-
x
ble, thus, the assumption is incorrect and ( S [ x ] ( p, f ))T p F , AS ( f ) , similar,
x
( S [ x ] ( p, f ))T , AS ( f ) p , what means | {S [ x ] ( p, f )} AS x ( f ){ p} .
| { p}id{ p} follows from the definition.
Let us prove | { p} f {q} | {q}g{r} pT , f qT | { p} f g{r} .
We have | { p} f {q},| {q}g{r} that means pT q F , f ; qT r F , g . We
need to show that pT r F , f g .
Let it be false and d : d pT d r F , f g . This means that
p(d ) T r ( f g (d )) F .
But using the additional condition we have pT , f qT , thus q( f (d )) T . That
means f (d ) qT , then f (d ) r F , g . This contradicts the fact that
F ,g
r ( f g (d )) F f g (d ) r f (d ) r
F
.
We have the contradiction, which means that the assumption is wrong and
pT r F , f g . Then | { p} f g{r} .
Let us prove | {b p} f {q} | {b p}g{q} | { p}IF (b, f , g ){q} .
We have | {b p} f {q}, | {b p}g{q} , which means:
(b p)T q F , f ;(b p)T q F , g .
We need to show that pT q F , IF (b, f , g ) .
Let d : d pT d q F , IF (b , f , g ) . Then p(d ) T , q( IF (b, f , g )(d )) F .
Let us examine different cases of b(d ) :
b(d ) is impossible, because then IF (b, f , g )(d ) leads to a contradiction with
assumptions about existence of such d .
b(d ) T IF (b, f , g )(d ) f (d ) d (b p)T IF (b, f , g )(d ) (b p )T , f .
With properties of the upper part of the inference rule we have:
(b p )T q F , f IF (b, f , g )(d ) f (d ) d (b p )T d q F ,( IF (b , f , g ) .
A case with b(d ) F is similar to the case where b(d ) T .
(b p)T q F , g IF (b, f , g )(d ) g (d ) d (b p)T d q F ,( IF (b, f , g ) .
Thus d q F ,( IF (b, f , g ) in any case if d is defined which is guaranteed by the as-
sumption. That leads us to the contradiction, so, pT q F , IF (b , f , g ) .
Thus, we have | { p}IF (b, f , g ){q} .
Let us prove | {b p} f { p} (b p)T , f pT | { p}WH (b, f ){b p} .
We have | {b p} f { p} , that means: (b p )T p F , f .
We need to show that the following condition holds: pT (b p) F ,WH (b , f ) .
Program Algebras with Monotone Floyd-Hoare Composition 547
Let d : d pT d (b p) F ,WH (b , f ) . Then d n : d n WH (b, f )(d ) , and by the
definition of the composition we have d1 , d 2 , d n : d d1 di 1 f (di ) ,
i 1, n 1 b(d j ) T , j 1, n 1 p(d1 ) T b(d n ) F and (b p)(d n ) F .
Thus, (b p)(d1 ) T . By (b p )T , f pT we obtain that p(d 2 ) p( f (d1 )) T .
Using the induction over a number of loop execution we obtain that p(d n ) T .
That means (b p)(d n ) T . Thus, we obtained contradiction, so,
F ,WH ( b , f )
p ( b p )
T
.
Let us prove | { p} f {q} pT pT q F qF | { p} f {q} .
We have | { p} f {q} , what means: p T q F , f .
We need to show that pT q F , f .
Let this condition be false and d : d pT d q F , f . This means that
p(d ) T q( f (d )) F .
By the condition that pT p T we obtain p(d ) T .
But pT q F , f , thus, d q F , f , while d q F , f , then f (d ) , and we
have q( f (d )) F .
We have a contradiction which means that the assumption does not hold,
so, pT q F , f .
Both conditions are proved, then | { p} f {q} .
Thus, all rules are inspected and theorem is proved.
Also the condition for the rule Ax_SEQ can be substituted by one of the following:
pT , f q , q , g r F or q F , g r F , but none of them is a sufficient condi-
tion, because (| { p} f {q} | {q}g{r} | { p} f g{r}) pT , f qT doesn’t hold.
Similar for the rule Ax_WH , the condition could be given in the one of the follow-
ing manner: (b p)T , f p , (b p) , f p F or (b p) F , f p F , and they
are also insufficient.
The conditions for the rule Ax_CONS also are not sufficient. To prove that we need
only to show an example when the condition does not hold but the rule does.
But in some cases we can avoid adding the conditions implicitly to the rules.
Theorem 4. For all assertions { p} f {q} that were inferred using rules of the infer-
ence system except Ax_CONS the following properties hold:
pT , f qT ,
pT , f q ,
p , f q F ,
pF , f qF .
Let us prove the first property by induction. For the fourth property the case is
similar and second and third properties are consequences of the first and the fourth
respectively.
548 A. Kryvolap, M. Nikitchenko and W. Schreiner
Induction base: for Ax_ID and Ax_AS proof is obvious.
Induction step. For Ax_SEQ we have:
pT , f q T q T , g r T pT , f g r T .
The proof of this fact is obvious.
For Ax_IF we need to prove:
(b p)T , f qT (b p)T , g (q)T pT , IF (b , f , g ) qT .
Consider d pT , IF (b , f , g ) , then x : ( IF (b, f , g )( x) d ) p( x) , that leads to two
cases:
– b( x) T , then f ( x) qT , moreover d IF (b, f , g )( x) f ( x) , thus, d qT ;
– b( x) F , then g ( x) qT , moreover d IF (b, f , g )( x) g ( x) , thus, d qT .
For Ax_WH we need to prove: (b p)T , f pT pT ,WH (b, f ) (b p)T .
Let d pT ,WH (b, f ) , then x : (WH (b, f )( x) d ) p( x) , we need to prove that
(b p)(d ) T . Let us examine all data that are obtained during the calculation of
WH (b, f )( x) : x x0 ; x1 f ( x0 ) ... d xn , b( x0 ) b( x1 ) b( xn 1 ) T ,
b( xn ) F , thus from (b p)T , f pT we have, that p(d ) p( xn ) T , this together
with b( xn ) F gives (b p)(d ) T , what was needed to prove.
The theorem is proved.
Theorem 3 and Theorem 4 together give us the fact that if we declare Ax_CONS in
such a way that it retains the properties of the theorem 4, then inference rules system
will be sound without addition of new conditions, which will be guaranteed by Theo-
rem 4.
But in this case system would not be complete. Let us give an example.
Let q be an arbitrary predicate that has nonempty truth, false, and undefiniteness
domains, and p be such predicate that pT qT q . Then | { p}id{q} , but
( pT ,id qT ) , when the inference rules system was constructed for the following
property to hold: | { p} f {q} pT , f qT .
5 Conclusions
In this paper special program algebras of partial quasiary mappings have been de-
scribed. Such algebras form a semantic base for a modified Floyd-Hoare logic. In this
case assertions have been presented by a special composition called Floyd-Hoare
composition. Monotonicity and continuity of this composition have been proved. The
language of the modified Floyd-Hoare logic has been described. Further, the inference
rules for such a logic have been studied and their soundness conditions have been
specified. The logic constructed can be used for program verification.
The major directions of further investigation are the question of completeness of
the system of inference rules, invariants for rules, and types for variables and func-
tions. Also the authors plan to construct a prototype of a program system in the style
of [8, 9] oriented on the constructed logics.
Program Algebras with Monotone Floyd-Hoare Composition 549
References
1. Floyd, R. W.: Assigning Meanings to Programs. In: Proc. American Mathematical Society
Symposia on Applied Mathematics, vol. 19, pp. 19–31 (1967)
2. Hoare, C. A. R.: An Axiomatic Basis for Computer Programming. Comm. ACM, 12, 576–
580, 583 (1969)
3. Nikitchenko, M. S., Shkilniak, S. S.: Mathematical Logic and Theory of Algorithms. Pub-
lishing house of Taras Shevchenko National University of Kyiv, Kyiv (2008) (in Ukrain-
ian)
4. Nikitchenko, M., Tymofieiev, V.: Satisfiability and Validity Problems in Many-Sorted
Composition-Nominative Pure Predicate Logics. In: V. Ermolayev et al. (eds.): ICTERI
2012, CCIS 347, pp. 89–110. Springer Verlag, Berlin Heidelberg (2013)
5. Nikitchenko, M. S., Tymofieiev, V. G.: Satisfiability in Composition-Nominative Logics.
Central European Journal of Computer Science, 2(3), 194–213 (2012)
6. Nielson. H.R., Nielson, F.: Semantics with Applications: A Formal Introduction. John
Wiley & Sons Inc. (1992)
7. Avron, A., Zamanskym A.: Non-Deterministic Semantics for Logical Systems. Handbook
of Philosophical Logic, vol. 16, pp. 227–304 (2011)
8. Schreiner, W.: Computer-Assisted Program Reasoning Based on a Relational Semantics of
Programs. In: P. Quaresma and R.-J. Back (eds.) Proc 1st Workshop on CTP Components
for Educational Software (THedu'11), July 31 2011, Wrocław, Poland, No 79 of Electronic
Proceedings in Theoretical Computer Science (EPTCS), ISSN: 2075-2180, pp. 124–142
(2012)
9. Schreiner, W.: A Program Calculus Technical Report. Research Institute for Symbolic
Computation (RISC), Johannes Kepler University, Linz, Austria, http://www.risc.uni-
linz.ac.at/people/schreine/papers/ProgramCalculus2008.pdf (2008)
A Formal Model of Resource Sharing Conflicts
in Multithreaded Java ?
Nadezhda Baklanova and Martin Strecker
Institut de Recherche en Informatique de Toulouse (IRIT), Université de Toulouse
{nadezhda.baklanova,martin.strecker}@irit.fr
Abstract. We present a tool for analysing resource sharing conflicts
in multithreaded Java programs. We consider two models of execution:
purely parallel one and sequential execution on a single processor. A Java
program is translated into a system of timed automata which is verified
by the model checker Uppaal. We also present our work in progress on
formalisation of Real-Time Java semantics and the semantics of timed
automata.
Keywords. resource sharing, Java, timed automata, model checking
Key terms. Model, Development, ConcurrentComputation, FormalMethod,
QualityAssuranceProcess
1 Introduction
Along with increasing usage of multithreaded programming, a strong need of
sound algorithms arises. The problem is even more important in programming
of embedded and real-time systems where liveness conditions are extremely im-
portant. To certify that no thread would starve or would be deadlocked, lock-free
and wait-free algorithms have been developed. Lock-free algorithms do not use
critical sections or locking and allow to avoid thread waiting for getting access to
a mutual exclusion object. Nevertheless, only one thread is guaranteed to make
progress. Wait-free algorithms prevent starvation by guaranteeing a stronger
property: all threads are guaranteed to make progress, eventually. Such algo-
rithms for linked lists, described for example in [4,11], are very complex, difficult
to implement and, consequently, hard to verify.
What is worse, these algorithms seem to be incompatible with hard real-time
requirements: the progress guarantees are not bounded in time. Thus, a lock-free
insertion of an element into a linked list by a thread may need several (possi-
bly infinitely many) retries because the thread can be disturbed by concurrent
threads. Under these conditions, it is not possible to predict how much time is
needed before the thread succeeds.
?
Part of this research has been supported by the project Verisync (ANR-10-BLAN-
0310)
A Formal Model of Resource Sharing Conflicts in Multithreaded Java 551
Critical sections are used in many applications in order to ensure concurrent
access to objects although if the scheduling order is wisely planned, locks are not
necessary since threads access objects at different moments of time. This is the
motivation for our work: we develop a tool for checking resource sharing conflicts
in concurrent Java programs based on the statement execution time. This gives a
“time-triggered” [6] flavor to our approach of concurrent system design: resource
access conflicts are resolved by temporal coordination at system assembly time,
rather than during runtime via locking or via retries (as in wait-free algorithms).
We assume that a program is annotated with WCET information known from
external sources. The checker translates a Java program into a timed automaton
which is then model checked by a tool for timed automata (concretely, Uppaal).
In this paper, after an informal introduction (Section 2), we present a formal
semantics of the components of the translation, namely Timed Automata (Sec-
tion 3) and a multi-threaded, timed version of Java (Section 4). Then we describe
the mechanism of the concrete translator written in OCaml (Section 5) and give
some preliminary correctness arguments (Section 6) – the formal verification still
remains to be done. The conclusions (Section 7) discuss some restrictions of our
current approach and possibilities to lift them.
Status of the present document: We give a glimpse at several aspects of our
formalisation, which is far from being coherent. Therefore, this paper is rather
a basis for discussion than a finished publication.
2 Informal Overview
To show the main idea, we present an example of a concurrent Java program.
It is a primitive producer-consumer buffer with one producer and one consumer
where both producer and consumer are invoked periodically. The program is
annotated with information about statement execution time in //@ ... @//
comments.
private class Run1 implements Runnable {
public void run () {
int value , i ;
// @ 1 @ //
i =0;
while (i <10) {
synchronized ( res ) {
// @ 2 @ //
value = Calendar . getInstance () . get ( Calendar .
MILLISECOND ) ;
// @ 5 @ //
res . set ( value ) ;
}
Thread . sleep (10) ;
// @ 2 @ //
i ++;
552 N. Baklanova and M. Strecker
}
}
}
private class Run2 implements Runnable {
public void run () {
int value , i ;
// @ 1 @ //
i =0;
Thread . sleep (9) ;
while (i <10) {
synchronized ( res ) {
// @ 4 @ //
value = res . get () ;
}
Thread . sleep (8) ;
// @ 1 @ //
i ++;
}
}
}
One of the possible executions is shown in Figure 1.
01 8 20
Fig. 1: Possible execution flow. Black areas represent execution without locks,
blue and green areas - execution within a critical section, grey areas - sleeping,
white areas - waiting for processor time.
After having translated this program to a system of timed automata we run
the Uppaal model checker to determine possible resource sharing conflicts. The
checked formula is
A[]∀(i : int[0, objN umber − 1])∀(j : int[0, autN umber − 1])waitSet[i][j] < 1,
(1)
where waitSet is an array of boolean flags indicating whether a thread waits
for a lock of a particular object. If all array members in all moments of time
are false, no thread waits for a lock therefore no resource sharing conflicts are
possible.
A Formal Model of Resource Sharing Conflicts in Multithreaded Java 553
3 Timed Automata Model
Timed automata are a common tool for verifying concurrent systems; the under-
lying theory is described in [1]. We formalize the basic semantics of an extension
of timed automata used in the Uppaal model checker. The formalized syntax
and semantics are adapted from [3].
An automaton edge is composed of a starting node, a condition under which
the edge may be taken (guard), an action, clocks to reset and a final node.
type-synonym ( 0n, 0a) edge = 0n × cconstr × 0a list × id set × 0n
An invariant is a condition on a node which must be satisfied when an au-
tomaton is in this node.
type-synonym 0n inv = 0n ⇒ cconstr
An automaton consists of a set of nodes, a starting node, a set of edges and
an invariant function.
type-synonym ( 0n, 0a) ta = 0n set × 0n × ( 0n, 0a) edge set × 0n inv
We use the model checker Uppaal which proposes an extension of classical
timed automata with variables. A full state comprises a node and a valuation
function. There are two types of variables: integer (aval ) and boolean (bval ),
and clock variables which have a special semantic status in timed automata.
The available transitions can be defined knowing given this state.
record valuation=
aval :: id ⇒nat
bval :: id ⇒bool
cval :: id ⇒time
type-synonym 0n state = 0n × valuation
A timed automaton can perform two types of transitions: timed delay and
edge taking. If an automaton takes an edge, variables may be updated or clocks
may be reset to 0. The Transition constructor takes a list of variable and clock
updates as an argument.
datatype 0a ta-action = Timestep time | Transition 0a list
Accordingly, the timed automata semantics has two rules: delay and tran-
sition. If an automaton is delayed, it stays in the same node, and values of all
clocks are increased to the value d. The invariant of the current node must be
satisfied.
If an automaton takes a transition, the node is changed, and clock values
remain unchanged unless they are reset to 0. The invariants of both starting
and final nodes must be satisfied as well as the guard of the taken transition. If
the action of the transition involves variable updates, the valuation function is
updated as well. This is done by the function application eval-stmts varv a.
554 N. Baklanova and M. Strecker
varv 0 = varv (|cval := add (cval varv ) d |)
varv |= invs l varv 0 |= invs l l ∈ nodes
(nodes, init, edges, invs) ` (l , varv ) −Timestep d → (l , varv 0)
(l , g, a, r , l 0) ∈ edges
varv |= g varv 0 = eval-stmts varv a(|cval := reset (cval varv ) r |)
varv 0 |= invs l 0 l ∈ nodes l 0 ∈ nodes
(nodes, init, edges, invs) ` (l , varv ) −Transition a→ (l 0, varv 0)
A network of timed automata is a product automaton with exceptions in case
of handshaking actions [2]. Handshaking allows to synchronize two automata so
that both take an edge simultaneously.
(s 1 , g 1 , a, cs 1 , s 1 0) ∈ edges 1 (s 2 , g 2 , a, cs 2 , s 2 0) ∈ edges 2
a ∈ getEdgeActions edges 1 ∩ getEdgeActions edges 2
((s 1 , s 2 ), g 1 d∧e g 2 , a, cs 1 ∪ cs 2 , s 1 0, s 2 0) ∈ edges-shaking edges 1 edges 2
(s 1 , g, a, cs, s 1 0) ∈ edges 1 a ∈ getEdgeActions edges 1
a ∈ / getEdgeActions edges 2 s 2 ∈ getEdgeNodes edges 2
((s 1 , s 2 ), g, a, cs, s 1 0, s 2 ) ∈ edges-shaking edges 1 edges 2
(s 2 , g, a, cs, s 2 0) ∈ edges 2 a ∈ getEdgeActions edges 2
a ∈ / getEdgeActions edges 1 s 1 ∈ getEdgeNodes edges 1
((s 1 , s 2 ), g, a, cs, s 1 , s 2 0) ∈ edges-shaking edges 1 edges 2
4 Java Model
The look of the Java semantics has been inspired by the Jinja project [5] and its
multithreaded extension JinjaThreads [7]. The Java execution flow is modeled
by three transition relations: evaluation, scheduler and platform. The evaluation
semantics is the semantics of a single thread, the scheduler semantics is respon-
sible for thread scheduling, and the platform semantics formalizes the notion
of time advancement. Taking into account the passage of time is the essential
increment wrt. the above-mentioned semantics.
Following Jinja, we do not distinguish expressions and statements; their
datatype is the following:
datatype expr
= Val val
| Var vname
| VarAssign vname expr
| Cond expr expr expr
| While expr expr
A Formal Model of Resource Sharing Conflicts in Multithreaded Java 555
| Annot annot expr
| Sync expr expr
| Sleep expr
... and others.
The system state is large and complex; it stores local information of all
threads such as local variable values and expression to be evaluated, shared
objects, time, actions to be carried out by the platform, locks and wait sets.
Given this state and scheduler logic, the further execution order is deterministic.
record full-state =
threads :: id ⇒ schedulable option — threads pool
th-info :: thread-id ⇒ thread-state option — expression to be evaluated and state
sc-info :: schedule-info — locks and thread statuses
pl-info :: platform-info — global time
gl-info :: heap — heap state
ws-info :: waitSet — wait sets
running-th :: thread-id — currently running thread
pending-act :: action — action to be carried out by platform
Evaluation semantics depends solely on local and heap variables therefore we
use the reduced variant of full state for thread evaluation step.
record eval-state =
ev-st-heap :: heap
ev-st-local :: locals
When a thread expression is reduced, the duration of the performed action
is not taken into account on the evaluation step, so the action type is passed
further to the platform step where time advances according to the action. For
now we assume that any action of a particular type takes a fixed amount of time
for execution.
The evaluation rules take an expression and a local state and translate them
to the new pair of expression and state and also emit an action for the platform
step. Here are some examples of evaluation rules.
fs ` (e, s) −act→ (e 0, s 0)
fs ` (VarAssign vr e, s) −act→ (VarAssign vr e 0, s 0)
fs ` (VarAssign vr (Val vl ), s) −EvalAct (exec-time VarAssignAct)→ (Val Unit,
s(|ev-st-local := ev-st-local s(vr 7→ vl )|))
Several rules use information about locks and wait sets that is not included
in the local state therefore they pull it from the full state.
¬ locked a fs
fs ` (Sync (Val (Addr a)) e, s) −lock-action a fs 0 → (Sync (Val (Addr a)) e, s)
556 N. Baklanova and M. Strecker
locked a fs fst (the (sc-lk-status (sc-info fs) a)) 6= runnint-th fs
fs ` (Sync (Val (Addr a)) e, s) −lock-action a fs 0 → (Throw [ResourceSharingConflict], s)
locked a fs
fs ` (Sync (Val (Addr a)) (Val v ), s) −unlock-action a fs 0 → (Val Unit, s)
5 Abstracting Java to Timed Automata
We consider two models of program execution. The first one is purely parallel, i.e.
each thread is assumed to execute on its own processor so that no thread waits
for CPU time. Another model is the sequential one when a program executes
on a single processor. The parallel model is easier, however, it does not seem to
be realistic. The sequential model represents the realistic situation for real-time
Java applications since the RTSJ (Real-Time Specification of Java [8]) specifies
the behavior for monoprocessor systems only.
The translated Java programs must be annotated with timing information
about execution time of the following statement. The translation uses timing
annotations to produce timed automata which model the program. The obtained
system is model checked for possible resource sharing conflicts.
5.1 General principles
We suppose that the translated program has a fixed number of threads and
shared fields, all of them defined statically. The initialization code for threads
and shared fields must be contained in the main method. The classes implement-
ing Runnable interface must be nested classes in the class containing the main
method. The required program structure is shown in the figure 2.
Each thread created in the program is translated into one automaton, and
one more additional automaton modeling the Java scheduler is added to the
generated system.
Java statements are translated into building blocks for condition statement,
loop etc. which are assembled to obtain the final automaton. Annotated state-
ment is translated into its own block. Method calls and wait/notify statements
are not translated for now.
The timed automata system contains an array of object monitors representing
acquired locks on shared objects. When a thread acquires a lock of an object,
the monitor corresponding to this object is incremented, and when the lock is
released, the monitor is decremented.
There is a number of checks which are performed before on the program
source code which guarantee correctness of the generated model. One of the most
critical is the requirement that the whole parse tree must be annotated, i.e. for
any leaf of the AST there is a timing annotation somewhere above this leaf.
With this requirement the behavior of the generated system can be determined
in each moment of time.
A Formal Model of Resource Sharing Conflicts in Multithreaded Java 557
public class Main {
Res1 field1 ; // shared fields declaration
public static void main ( String [] args ) {
Run1 r1 ; // declarations of Runnable object instances
Thread t1 , t2 ; // thread declarations
r1 = new Run1 () ; // Runnable objects initialization
field1 = new Res1 () ; // shared fields initialization
t1 = new Thread ( null , r1 , " t1 " ) ; // thread creation
t1 . start () ; // thread start
}
private class Run1 implements Runnable {
public void run () { // thread logic implementation
...
}
}
}
private class Res1 { // resouce classes
...
}
Fig. 2: Required program structure
5.2 Parallel model
In the parallel model threads are supposed not to wait for processor time if
they want to execute a statement. However, threads can wait for a lock if they
need one which has been taken by another thread. Since this model is not very
realistic we concentrate on the sequential model further.
5.3 Sequential model
In the sequential model we assume that at every moment of time only one thread
or scheduler can execute. Threads which do not execute in a particular moment
of time wait for processor time. Also threads can wait for a lock; waiting does
not consume CPU time.
Automata communicate with the scheduler through channels: if the scheduler
has selected one thread, it sends a message to it so the thread starts executing.
After finishing its execution, the thread sends a message to the scheduler, and
the next scheduling cycle starts. The scheduler uses channels run[i] to call the
i-th automaton, and the automata use the channel scheduler to give the control
back to the scheduler.
There is an array of clocks c[i], each of them corresponding to one thread
automaton. These clocks are used to calculate time of annotated statements
execution or sleeping time. There is also one clock cGlobal used for tracking
global time.
558 N. Baklanova and M. Strecker
START start1 START
run[i]? run[i]?
execTime[i]=,
c[i]=0
final1_start2
MIDDLE
MIDDLE
c[i]<=execTime[i]
c[i]>=execTime[i]
schedule!
schedule!
execTime[i]=0
FINAL final2 FINAL
(a) Assignment (b) Sequence (c) Annotation
auxIf start1 auxIf start1
schedule! schedule!
run[i]? run[i]?
START final1 START final1_final2
run[i]? run[i]?
schedule! schedule!
auxElse auxElse start2
(d) Condition I (e) Condition II
Fig. 3: Building blocks for automata. Elements added on the current step are
red; blue and green elements have been generated in the previous step.
The building blocks and their translation are the following for the sequential
model:
(a) Assignment (3a). Three new states and two transitions between them are
added. The transition from START to MIDDLE listens to the channel run[i],
and the transition from MIDDLE to FINAL calls the channel schedule. The
state MIDDLE is urgent since we assume that any statement except the an-
notated one takes time for execution.
(b) Sequence (3b). Having two automata with start and final states called start1,
start2 and final1, final2 correspondingly, the states final1 and start2
are merged.
(c) Annotation (3c). Three new states and two transition between them are
added. The transition from START to MIDDLE listens to the channel run[i],
sets the variable execTime[i] to the value of the current annotation and
resets the clock c[i] to 0. The transition from MIDDLE to FINAL calls the
channel schedule and resets the variable execTime[i] back to 0. The state
MIDDLE has an invariant forbidding the automaton to stay in this state if the
value of the clock c[i] bypasses execTime[i]. The transition from MIDDLE
A Formal Model of Resource Sharing Conflicts in Multithreaded Java 559
start1 final1
run[i]?
schedule!
_monitor[j]--
schedule!
auxIn auxOut auxLoop
schedule!
_monitor[j]==0 run[i]?
run[i]? start1
_monitor[j]++,
waitSet[j][i]=0
FINAL
START
START
schedule! run[i]?
auxWait
_monitor[j]>0 FINAL
run[i]? auxEnd
waitSet[j][i]=true schedule!
(a) Lock (b) Loop
MIDDLE
c[i]>=execTime[i]
schedule! run[i]?
execTime[i]=0
START auxSleep auxWake FINAL
run[i]? schedule!
execTime[i]=,
c[i]=0
(c) Sleeping
Fig. 4: Building blocks for automata. Elements added on the current step are
red; blue and green elements have been generated in the previous step.
to FINAL has a guard enabling this transition only if the value of c[i] is
greater or equal to execTime[i]. The invariant and the guard ensure that
the automaton would be in the MIDDLE state as long as the annotation claims.
(d) Condition (3d,3e). Three new states and four transitions are added. If both
if and else branches are presented, the final states of automata representing
the branch internals are merged. The states auxIf and auxElse are auxiliary
states introduced to divide listening and calling transitions therefore they
are made committed. Two transitions from START to auxIf and auxElse
listen to the channel run[i], and the transitions from auxIf to start1 and
from auxElse to start2 (or to final1 in case of absence of the else branch)
call the channel schedule.
(e) Lock (4a). Two meaningful states and two auxiliary states are added. The
transition from START listens to the channel run[i] and has a guard checking
whether a lock for the object in the argument of the synchronized statement
is not taken by other threads.
560 N. Baklanova and M. Strecker
(f) Loop (4b). Two meaningful states and two auxiliary states are added. One
transition from the START goes to the next loop iteration, another one exits
the loop. Both transitions from START to auxLoop and auxEnd listen to the
channel run[i]. The transitions from auxLoop to start1 and from auxEnd
to FINAL call the channel schedule. Both auxLoop and auxEnd are made
committed. The final state of the automaton corresponding to the loop body
is merged with the START state.
(g) Sleeping (4c). The automaton for sleep statement resembles the automa-
ton for annotated statement with additional elements for returning control
to the scheduler during sleeping. There are three meaningful and two aux-
iliary states with transitions connecting them into a chain. The auxiliary
states, auxSleep and auxWake, are committed. The transition from START
to auxSleep listens to the channel run[i], sets the variable execTime[i] to
the duration of sleeping period and resets the clock c[i] to 0. The transition
from auxSleep to MIDDLE calls the channel schedule so that the scheduler
can schedule other threads. The transition from MIDDLE to auxWake listens
to the channel run[i] and has a guard enabling this transition only if c[i]
is greater or equal to execTime[i]. The update on this channel resets the
value of execTime[i] back to 0. Unlike the automaton for the annotated
statement, there is no invariant in the MIDDLE state because a thread is not
obliged to continue its execution right after it has woken up. It may wait
for processor time before. The transition from auxWake to FINAL calls the
schedule channel.
5.4 Scheduler
The Java scheduler maintains thread statuses and grants permission to execute
to threads. The scheduler model has three states: scheduling, runThread, wait.
The scheduler starts in the state scheduling which has transitions for updating
thread eligibility statuses. When all thread statuses are updated, the scheduler
moves to the state runThread calling the channel run[i] for some thread with
index i which is eligible for execution. While the thread is executing, the sched-
uler stays in the state runThread. When the thread has finished its execution,
it calls a channel schedule, and the scheduler returns back to the scheduling
state, and the new scheduling cycle starts. If there was no thread eligible for
execution, the scheduler goes to the wait state where it can stay for some time
and repeat scheduling.
Each thread gets two transitions for status updates. One assumes that a dead-
line for an action performing by a thread has passed, another one assumes that
the deadline has not been reached yet. In the first case the flag isEligible[i]
is set to true, and the thread with index i can be scheduled for execution. Oth-
erwise, isEligible[i] is set to false, and the thread with index i cannot be
scheduled.
A Formal Model of Resource Sharing Conflicts in Multithreaded Java 561
c[0]>=execTime[0]&&!isUpdated[0] c[1]>=execTime[1]&&!isUpdated[1]
isEligible[0]=true, isEligible[1]=true,
isUpdated[0]=true, isUpdated[1]=true,
updateAllStatuses() updateAllStatuses()
c[0]. A mapping from variables to sets of possible values is a domain. Some popular
domains for constraint programming are:
Boolean domains, where only true/false constraints apply (SAT problem)
Integer domains, rational domains
Linear domains, where only linear functions are described and analyzed (although
approaches to non-linear problems do exist)
Finite domains, where constraints are defined over finite sets
Mixed domains, involving two or more of the above
Finite domains are one of the most successful domains of constraint programming.
In some areas (like operation research), constraint programming is often identified
with constraint programming over finite domains.
Definition 2 A domain d is a function mapping variables to sets of values, such
that d(x) V. The set of all domains is Dom: = X → P (V). The set of values in d for
a particular variable x, d(x), is called the variable domain of x. A domain d represents
a set of assignments, a constraint, defined as
con(d) := a Asn | x X : a(x) d(x) (3)
Said that an assignment a con(d) is licensed by d.
In our example, we can implement two types of domain realization. Each domain
can be realized as a state of an agent, and be (or not) omitted by propagator during
insertion, or other way – store all sets of domain in environment' state.
Definition 3 A constraint satisfaction problem (CSP) is a pair of a domain d
and a set of constraints C. The constraints C are interpreted as a conjunction of all c Є
C and are thus equivalent to the constraint a Asn | c C : a c. The solutions of
a CSP are the assignments licensed by d that satisfy all constraints in C,
defined as
sol(< d,C >) := a con(d) | c C : a c )
2 Propagators
The basis of a propagation-based constraint solver is a search procedure, which sys-
tematically enumerates the assignments licensed by the domain d of a CSP .
Implementation of Propagation-Based Constraint Solver in IMS 569
For each assignment, the solver uses a decision procedure for each constraint to de-
termine whether the assignment is a solution of the CSP. Enumerating all assignments
would be infeasible in practice, so in addition to the decision procedure, the solver
employs a pruning procedure for each constraint, which may rule out assignments
that are not solutions of the constraint.
Two problems, decision, and pruning procedure for constraints implemented by
propagators. Each propagator induces particular constraint. Propagator decides for a
given assignment, whether it satisfies the induced constraint, and it can cut off (prune)
these tasks from the domain that do not satisfy the constraint. Interleaving propaga-
tion and search yield sound and complete procedure for solving CSP. It is complete,
because only the assignments that are not solutions are pruned by propagators, and all
other assignments are in enum. This is sound, because for each of these tasks, the
propagators to decide whether it's the definition of solution. The formal definition of
propagators author (see [1]), reflects the minimum properties that are needed in order
to get a sound and complete solver. Thus, this model differs from the one commonly
found in the literature. Furthermore, knowledge of the unique characteristics of the
propagators induced constraints is new. The authors define the propagators in terms
domains.
A propagator is a function p that takes a domain as its argument and returns a
stronger domain, it may only prune assignments. If the original domain was an as-
signed domain {a}, the propagator either accepts it (p({a}) = {a}) or rejects it (p({a})
= 0), realizing the decision procedure for its constraint. In fact, each propagator in-
duces a unique constraint, the set of assignments that it accepts. To make this setup
work, we need one additional restriction. The decision procedure and the pruning
procedure must be consistent: if the decision procedure accepts an assignment, the
pruning procedure must never remove this assignment from any domain—this prop-
erty is called soundness.
Definition 4 A propagator is a function p Dom Dom that is:
Contracting: p(d) d for any domain d
Sound: for any domain d Dom and any assignment a Asn , if a d , then
p( a ) p(d)
The set of all propagators is Prop. If a propagator p returns a strictly stronger do-
main (p(d) d) , we say that p prunes the domain d. The propagator p induces the
constraint c p defined by the set of assignments accepted by p:
c p := a Asn | p( a) = a (5)
Soundness expresses exactly that the decision and the pruning procedure realized
by a propagator are consistent. A direct consequence is that a propagator never re-
moves assignments that satisfy its induced constraint.
Focusing on our problem, we implement the idea of propagators in additional func-
tions that will proceed the domains (as agent state or environment state) before inser-
tion. Then after the insertion call other propagators to prune from their induced do-
main unnecessary values to decreasing with each step the search field.
570 I. Ol. Blinov
For abstracting the solution of the problem we should give the definition and de-
scribing of propagation problem, as a higher model of solution of problems of given
type.
2.1 Propagation Problem
Propagators were defined as a refinement of constraints – each propagator induces
one particular constraint, but in addition has an operational meaning, its pruning pro-
cedure. It is possible to define the operational equivalent of a CSP, a propagation
problem. Propagation problems realize all constraints of a CSP using propagators.
Definition 5 A propagation problem (PP) is a pair of a domain d and a set
of propagators P. The induced constraint satisfaction problem of a propagation prob-
lem is the CSP d, c p | p P > . The solutions of a PP are the solu-
tions of the induced CSP, sol(< d, P >) := sol(< d, c p | p P >) .
The set of solutions of a PP d, P can be defined equivalently as
sol(< d, P >) := a Asn | p P : p( a ) = a , just applying the definitions of in-
duced constraints and solutions of CSPs.
Solution of propagation problem make by using propagators, at each step of insert-
ing an agent into environment. For this, as we mentioned earlier, we will inspect by
propagator each domain that is stored in the attributes of the agent. Before each inser-
tion, a domain stored in the attributes of the agent will checked by parameters gained
while working. Let’s we look at insertion machine architecture
Fig. 1. Insertion machine architecture
The calculation of a insertion function can be tested for the ability to insert the
propagator of the agent, which is inserted. If propagator exhausted domain that store
in the agent, inserting step will rejected.
Existence of strongest and weakest propagators. Propagators combine a deci-
sion procedure with a pruning procedure. While the decision procedure determines
the constraint a propagator induces, there is some liberty in the definition of the prun-
Implementation of Propagation-Based Constraint Solver in IMS 571
ing, as long as it is sound. Thus, there are different propagators for the same con-
straint, and they can be arranged in a partial order according to their strength:
Definition 6 Let p1 and p2 be two propagators that induce the same constraint.
Then p1 is stronger than p2 (written p1 p2 ) if and only if for all domains
d, p1(d) p2 (d) .
2.2 Propagation as a Transition System
Propagation as a Transition System. A propagation-based solver interleaves con-
straint propagation and search, where constraint propagation means to prune the do-
main as much as possible using propagators, before search resorts to enumerating the
assignments in the domain. Propagating as much as possible means, in the context of
propagation problems, to compute a mutual fixed point of all propagators.
Transitions. Let be a propagation problem. If there is a propagator
p P that can prune the domain d, that is, if p(d) d , then applying p yields a new,
simpler propagation problem, . Soundness of p makes sure that the new
problem has the same set of solutions as the original problem,
sol(< d, P >) = sol(< p(d), P >) .
A propagation problem thus induces a transition system, where a transition is pos-
sible from a domain d to a domain d' d if there is a propagator p P such that p(d)
= d′ . Written such a transition
d | p d' (6)
Definition 7 Let d be a domain. A transition d | p d' with a propagator p to a
domain d′ is possible if and only if d′ = p(d) and d' d . The transition system of a
propagation problem consists of all the transitions that are possible with
propagators p P , starting from d. A terminal domain, that is, a domain d such that
there is no transition d | p p(d) for any propagator p P , is called stable.
Written d d' if there is a sequence of transitions that transforms d into a stable
domain d′. This sequence is empty, d d , if d is stable.
The transition system of a propagation problem is non-deterministic, as there are
many possible chains of propagation that result in a stable domain.
Fixed points. The important theorem that ensures that constraint propagation is
useful in practice is that, given a propagation problem , its transition system is
finite and terminating. No matter in what order the propagators are applied, we reach
a stable propagation problem after a finite number of steps.
The naive approach to solving a propagation problem is to generate all as-
signments a d , and then use the propagators p P to check whether a satisfies all
constraints. This approach makes use of the fact that propagators realize decision
procedures for their induced constraints, but does not use their pruning capabilities. A
solver that proceeds naively in this fashion is said to follow the generate-and-test
approach.
572 I. Ol. Blinov
3 Conclusion
In conclusion, it can be said that the search for solutions by the generate-and-test
approach is inefficient, so we will consider other options. Nevertheless, this option
works well for prototyping, because of ease of implementation. In the future, we plan
to create a working prototype of the university schedule, which plans to make a uni-
versal, independent of the input parameters, the types of activities and a list of les-
sons. The most effective solution to this problem now is supposed to use multi-layer
environments, for pruning each of input domains by few environments and few
propagators.
References
1. Tack, G.: Constraint Propagation. Models, Techniques, Implementation. Saarbrucken
(2009)
2. Letichevsky, A., Letichevskyi, O., Peschanenko, V., Blinov, I., Klionov, D.: Insertion
Modeling System and Constraint Programming. In: Ermolayev, V. et al. (eds.) Proc. 7th
Int. Conf. ICTERI 2011, Kherson, Ukraine, CEUR-WS vol. 716 (2011)
3. Gilbert, D.R., Letichevsky, A.A.: A Universal Interpreter for Nondeterministic Concurrent
Programming Languages. Fifth Compulog Network Area Meeting on Language Design
and Semantic Analysis Methods (1996)
4. Letichevsky, A., Gilbert, D.: A General Theory of Action languages. Cybernetics and Sys-
tem Analyses, l(1), 16–36 (1998)
5. Letichevsky, A., Gilbert, D.: A Model for Interaction of Agents and Environments. In: D.
Bert, C. Choppy, P. Moses, (eds.): Recent Trends in Algebraic Development Techniques.
LNCS 1827, pp. 311–328, Springer Verlag, Berlin Heidelberg (1999)
6. Letichevsky, A.: Algebra of Behavior Transformations and its Applications. In:
Kudryavtsev, V.B., Rosenberg, I.G. (eds) Structural theory of Automata, Semigroups, and
Universal Algebra, NATO Science Series II. Mathematics, Physics and Chemistry,
Springer , vol. 207, pp. 241–272 (2005)
7. Martin, G., Selic, B. (eds.): UML for Real: Design of Embedded Real-Time Systems.
Kluwer Academic Publishers, Amsterdam (2003)
8. Letichevsky, A., Kapitonova, J., Letichevsky, A. Jr., Volkov, V., Baranov, S., Kotlyarov,
V., Weigert, T.: Basic Protocols, Message Sequence Charts, and the Verification of Re-
quirements Specifications. Computer Networks, 47, 662–675 (2005)
9. Kapitonova, J., Letichevsky, A., Volkov, V., Weigert, T.: Validation of Embedded Sys-
tems. In: Zurawski, R. (ed.) The Embedded Systems Handbook, CRC Press, Miami (2005)
10. Letichevsky, A., Kapitonova, J., Volkov, V., Letichevsky, A. Jr., Baranov, S., Kotlyarov,
V., Weigert, T.: System Specification with Basic Protocols. Cybernetics and System
Analyses, 4, 479–493 (2005)
UniTESK: Component Model Based Testing
Alexander K. Petrenko1, Victor Kuliamin1 and Andrey Maksimov1
1
Institute for System Programming of Russian Academy of Sciences
(petrenko, kuliamin, andrew)@ispras.ru
Abstract. UniTESK is a testing technology based on formal models or formal
specifications of requirements to the behavior of software and hardware com-
ponents. The most significant applications of UniTESK in industrial projects
are described, the experience is summarized, and the prospective directions to
the Component Model Based Testing development are estimated.
Keywords. Specification, verification, model-based testing, automated test
generation
Key terms. SoftwareSystem, FormalMethod, SpecificationProcess, Verifica-
tionProcess
1 Introduction
Model Based testing (MBT) is a rapidly developing domain of software engineering.
One of the reasons for such rapid development is the fact that MBT is at the intersec-
tion of various other domains of software engineering. In particular, those domains
include methods for defining, formalization and modeling of requirements, methods
for analysis of both formal specification and formal models as well as the software
code, methods for abstraction level control, model transformation and many other
software engineering domains. It provides MBT with the ability to quickly adopt the
recent achievements proved to be useful in joint domains, in particular, in methods of
static analysis and mixed static/dynamic analysis. However, there is no market-ready
well-established MBT tool that could be recommended for use in wide range of soft-
ware development and testing projects. To further develop MBT, we should first ana-
lyze the experience gained in the last 15-20 years in this domain. This should help to
identify some common problems and focus on their solutions. In this paper, we briefly
describe the stages of UniTESK (Unified TEsting & Specification toolKit) develop-
ment – one of the first MBT tools targeted at testing of wide class of software sys-
tems. In the course of the paper, we highlight both positive and negative lessons
learned during development and using UniTESK tools. The paper has a subtitle –
industrial paper. It means that here we don’t reveal any new solutions and don’t set
574 A. K. Petrenko, V. Kuliamin and A. Maksimov
any new scientific problems. We analyze the experience and try to learn lessons that
would be useful to researchers working in this domain.
The UniTESK technology [1, 2] was initially developed on the basis of the experi-
ence gained in the project on the creation of the automated testing KVEST [3] system,
which was developed for testing of the real-time operating system kernel. The work
started in 1994 when the term Model Based Testing did not exist. This term appeared
at the edge of the 21st century. Currently, MBT is rapidly developed. There are many
enthusiasts of this approach and many interpretations of the term itself. To properly
position UniTESK in the wide spectrum of MBT solutions, we should first clarify the
meaning of MBT within the UniTESK framework.
The following definition is currently given in Wikipedia: “Model-based testing is
application of Model based design for designing and optionally also executing arti-
facts to perform software testing. Models can be used to represent the desired behav-
ior of a System Under Test (SUT), or to represent testing strategies and a test envi-
ronment”.
This definition includes almost all known interpretations of this term. However,
most researchers and practitioners mean more specific approaches and testing tech-
niques by MBT. The first main dividing line is the choice of the modeled object:
some model the behavior of the target system (SUT), others model the environment of
the target system, in particular, the test itself or the testing system, which, of course,
are external to the target system. In UniTESK, the model specifies the system behav-
ior. There are also various types of MBT in this approach, which differ in the way of
the behavior description. About the first type Jan Paleska [4] says: “the behavior of
the system under test (SUT) is specified by a model elaborated in the same style as a
model serving for development purposes”. Such specifications or models are called
executable. The role of the executable model can be played either by prototype im-
plementation algorithm or by some model which explicitly contains the notion of
calculation/execution, for example, finite-state machine, Petri net, ASM [5], etc. Ex-
amples of other types of MBT, i.e. “nonexecutable”, are algebraic specifications,
software contracts in the form of pre- and post-conditions of functions. Each of the
model types has its own advantages and drawbacks when testing different SUTs. Be-
sides, when generating tests, not only test data and the sequence of calls of the tested
functions should be generated. Also required are the “artifacts” mentioned in the
Wikipedia definition, for example, test oracles – the components of the test suite that
automatically evaluate the results of the SUT execution whether they meet the re-
quirements or not. Executable models often do not allow test oracle creation, but they
are very good for generation of test sequences. Software contracts simplify generation
of test oracles, but they do not allow effective generation of test sequences. In other
words, several types of models required for generation of effective test suites. Back to
UniTESK, we can say that the main model type in it is the software contract in the
form of pre- and post-conditions of the functions. In addition, online construction of
finite-state machine is used in UniTESK making possible the generation of rather
non-trivial test sequences.
UniTESK is a technology that can be implemented on various software platforms
and, therefore, can be used for testing of API in various programming languages.
UniTESK: Component Model Based Testing 575
Currently, the most actively used implementations of UniTESK are for C, C++, Java,
Python. The corresponding tools are: CTESK, C++TESK, JavaTESK, and
PyTESK [6].
UniTESK is an academic product developed in ISPRAS. The UniTESK tools are
available under the free license. Experience of industrial application of UniTESK is
fused into the tools. Some test suites developed with UniTESK are included in official
test suites for certification of industrial software. For example, the OLVER test
suite [7] is one of the biggest MBT test suites in the world and yields to the only test
suite developed within the framework of the Microsoft Interoperability Initiative [8].
2 UniTESK Usage Review
Let’s consider the most interesting examples of UniTESK application and experience
gained in them.
The first application of UniTESK was in the project supported by Microsoft Re-
search on development of the MBT test suite for IPv6 implementation [9]. The project
started in 2000. At that time UniTESK was at the beginning of its development, so a
simplified (light) implementation of API testing in C was used – CTESK-light. In
spite of the tool instability, it allowed creation of the effective test suite that detected
defects, which were not detected by other test suites. It was the first experience of
using contract specifications for telecommunication protocol testing. It was demon-
strated that contract specifications in combination with the technique of testing sys-
tems with asynchronous interfaces developed within UniTESK [10, 11] allow creating
effective tests (they detected more defects, consumed less space and required less
effort for development and maintenance than tests developed with traditional tech-
nologies). However, the experience of protocols testing showed that besides the post-
conditions of the functions in the form of predicates it is useful to have executable
models when testing protocols.
One of the first experiences of using UniTESK implementation for testing Java
API [12] was the project on testing of Java run-time infrastructure developed as an
alternative to the popular Java-platforms. The development of the models and the tests
was not a problem since the interfaces were well documented. In addition to Java
interfaces, the target system contained also the interfaces in C++, but they also were
not a big problem since UniTESK architecture provides the layer of mediators-
adapters. The problems revealed when the actual testing started. MBT test suite with
online test generation is a fairly complicated program that has strong requirements to
the execution platform. In this case, the SUT itself was the execution platform which
still was not stable at that time. As a result, the test suite indicated the presence of
defects “everywhere”, which, in turn, was of no help to the developers.
The significant application of UniTESK on Java platform (JavaTESK) is the pro-
ject on testing of infrastructure of the distributed information system of one of Rus-
sian major mobile telephony provider. This project is still in progress. The possibility
of formal and rigorous specification of the components interfaces became the main
advantage of UniTESK for the customer in comparison with the other tools for func-
576 A. K. Petrenko, V. Kuliamin and A. Maksimov
tional testing. Hundreds of components were formally specified and tested with
UniTESK. By the end of the first year of using UniTESK the positive effect appeared
in shorter time of integration of new versions of the distributed system. However, a
serious problem revealed. In the previous UniTESK applications the requirements to
most interfaces were defined by standards and other well-developed documents. But
here the level of components documentation often appeared to be insufficient for
creation of consistent specifications. Recovery of documentation or requirements to
interfaces in the systems of such size becomes almost unsolvable task, which often
makes it impossible to use MBT in corpore. Possible solution of this problem will be
briefly discussed in Conclusion.
The largest example of UniTESK application is the OLVER (Open Linux VERifi-
cation) project [7] fulfilled in 2005-2007 under support of the Russian Ministry of
Education and Science. The goal of the project was to create formal specifications of
interfaces defined in the Linux Standard Base (LSB) standard or in LSB Core – the
central part of this standard, to be more exact. The LSB Core includes the most im-
portant libraries of OS Linux which implement most of the POSIX standard. The
rigorous description of the LSB standard and the test suite capable of a high-quality
checking of conformance of any Linux library implementation to the requirements of
the standard is a very powerful tool for providing portability of OS Linux applications
from one Linux distribution to another. The portability problem is very critical in the
Linux ecosystem, since several hundreds of very different distributions are available.
The project results are open [7]. The contract specifications of more than 1500 inter-
faces in C were created. Naturally, the CTESK tool was used for modeling and test
generation. In this project, the problems in the standards were also revealed: in LSB
(ISO/IEC 23360) and in The Single UNIX Specification containing the POSIX.1
standard (aka IEEE Std 1003.1, aka ISO/IEC 9945, aka The Open Group Base Speci-
fications Issue 6) as its significant part. The developed test suite is included into the
package of the certification tests of the international consortium The Linux Founda-
tion [13].
The experience of interface formalization for a large industrial standard and test
suite development for such standard gave many lessons to learn. One of such lessons
is importance of informational and methodological organization of such project. The
amount of documentation and sources, especially with respect to multiple versions
and variants for different hardware platforms, is huge. Besides, the development of
the standard and development of interface implementations involve thousands of peo-
ple around the world. It means that the documentation maintenance and availability is
one of the most important concerns of the projects of such scale. On the organiza-
tional and methodical side, we faced the fact that the training of new employees and
the specification and tests quality control require a lot of effort, and the quick
achievement of the required professional level is still impossible. In other words, the
scalability of the MBT projects in the part of increasing the number of specification
and verification experts is one of the most complicated problems preventing MBT
from wide introduction.
One of the methodical problems is the choice of the abstraction level for the model.
More abstract models or models separated into two-three layers of different abstrac-
UniTESK: Component Model Based Testing 577
tion levels simplify the reuse of the models and tests yielding, however, the bigger
and more complex test system. In the long term, it’s better to have multilayer models,
while in the short term the models close to implementation in the detail level (of
course, if the implementation already exists) are more appropriate. A professional and
experienced verification expert can find the balance between the abstract description
of the behavior, for example, of a file system and specifics and details of the interface
of its particular implementation. UniTESK provides special support for the separation
of abstraction levels. In particular, the specifics of interfaces can be encapsulated in
the mediator-adapter layer. The choice of the balance is determined by the long-tem
plans on using and improvement of the models and the test suite. So, the work of such
kind requires a broad experience and long-term planning skills, which can hardly be
expected from ordinary test engineers.
The results of the OLVER project were used later on in the development of the test
suite for the Russian real-time operating system OS2000/3000 [14]. This system pro-
vides two groups of interfaces. The first group meets the requirements of the POSIX
standard, the second one – the requirements of the ARINC-653 international standard
for the embedded and other safety critical systems. The definition of the adapter layer
separating model and implementation representations of the interfaces provided by
the UniTESK architecture significantly simplified the OLVER reuse in this project.
Along with the start of the OLVER project, the work on the UniTESK application
to testing of microprocessor designs [15] has been started. Hardware units being parts
of Russian microprocessors with the MIPS architecture and microprocessors with
VLIM/EPIC elements became the systems under test in this case. The size of typical
units in such microprocessors is several millions of gates. The tools required no sig-
nificant modifications for specification and test generation since CTESK was used as
the basis. Technically, binding CTESK to corresponding API of microprocessor
model simulator is not a problem, because most simulators that work with modeling
languages for microprocessors logic (HLD – High Level Design languages), for ex-
ample, VHDL or Verilog, provide suitable interface to C programs. Pre-conditions
semantics in contract specifications had to be slightly modified. They now describe
not just the domain of input data, but rather the operation execution readiness condi-
tions in the given time frame. The same as in the case of protocols modeling, the use
of explicit models of the target device behavior (functionality) along with the post-
conditions in the form of predicates appeared to be necessary.
Similar to the projects on verification of software systems, one of the main prob-
lems preventing MBT from introduction into practice (as well as many other verifica-
tion methods) is the lack of documentation and other descriptions of functional re-
quirements to components. However, the situation in microprocessors development is
slightly better than in the case of software development, because in this case it is cus-
tomary to build system and architectural models of instruction set semantics along
with the HLD models. Elements of these architectural models can be used to fill the
gap in the knowledge on behavior of some microprocessor design units [16]. It ap-
pears also relatively simple to implement parallel test execution on clusters. Typical
size of the finite-state machine generated during test execution for one microprocessor
unit is millions of nodes and dozens of millions of transitions. The algorithm of FSM
578 A. K. Petrenko, V. Kuliamin and A. Maksimov
generation and exploration on clusters with up to 200 nodes appeared to require just
10-15% overhead, i.e. scalability coefficient is close to 1.
It is important to mention the verification tasks that, on the one hand, could not be
reduced to modeling with contract specifications, and, on the other hand, pushed for-
ward the development of new MBT methods. In the first place, the task of compiler
testing should be mentioned, as well as the task of testing microprocessor as a whole,
the so-called “core testing”. The both cases are the tasks of system testing, where test
data and test stimuli are submitted to a big “black box” (in our case, these are test
programs submitted to the compiler or loaded into memory of the microprocessor
simulator), and it is interesting to test not just everything, but some specific behavior
modes or specific group of units. In the case of compiler testing, the OTK tool has
been developed that was used for testing of optimizing Intel compilers and Simulink
[17, 18]. It allows targeting on specific kinds of optimizations. In the case of micro-
processor design verification, the MicroTESK tool [19, 20, 21] was developed. The
main goal of this tool is checking of various situations appearing in the most compli-
cated subsystems of memory control: TLB, cache and Memory Management Unit
(MMU) as a whole.
3 Conclusions and Further Work
Let’s start with positive conclusions.
3.1 Positive Conclusions on Modern State of Using MBT
The world experience is confirmed [22], MBT can be effectively used in industrial
projects, and in comparison with the traditional testing MBT gives a unique advan-
tage – many defects can be found in requirements, which are often much more ex-
pensive than the defects in implementation
The achievable level of test coverage is significantly higher than the traditional one
(even in comparison with the “white box” testing). Thus, in the case of using OTK
for testing GCC compiler, the achieved test coverage was 95%, and in the case of
the Intel compiler this level was 75% that was significantly higher than the level
achieved by traditional tests [18].
Although the multi-level structure of specifications (several levels of abstraction) is
seldom used in practice, the explicit separation of adapters layer simplifies tests
porting and maintenance and, vise versa, the lack of the corresponding level of ad-
aptation makes test suite development significantly more complicated, which was
demonstrated in the Microsoft Interoperability Initiative program [23]
Online generation of test sequences with the FSM exploration method can be effi-
ciently parallelized and allows using computational resources of clusters with just
10-15% overhead, at least in the case of microprocessor models testing
The demand of MBT in the safety critical area increases. This tendency can be
found in standards defining requirements to development processes for safety criti-
cal systems, for example, in DO178C [24] and in Common Criteria [25].
UniTESK: Component Model Based Testing 579
3.2 Negative Aspects of the Modern State in the MBT Area
The main obstacle preventing MBT from wide introduction into practice is the
absence of specifications/models in casual software development. That is, the lack
of specifications is often not only the consequence of insufficient attention to
specification development or the consequence of short resources. The main reason
is often the lack of qualified specialists who are experts in the knowledge domain
and at the same time can create specification/model necessary for test generation.
If MBT is used in projects that do not involve MDD (Model Driven Development)
approach, then the model development delays the appearance of first tests – this
does not allow obtain tests early in the development. If MBT is used within MDD,
then the problems still remain, because different models required for development
and for testing, in particular, for generation of different artifacts of the test suite. It
is often considered as unacceptable additional cost, while with proper planning
many components of the development models can be reused during test generation
as demonstrated, for example, in M. M. Chupilko paper [16].
Bilingual test generation systems like UniTESK and first versions of SpecEx-
plorer [26], specification notations even close to conventional programming lan-
guages, for example, JML [27] make deployment of such systems difficult. Bilin-
gual notations require special training of the staff and need permanent and expen-
sive maintenance. Still note that the modern object-oriented languages already have
advanced means for writing specifications just in the same language [26-31].
3.3 Directions of Further Works
A variety of modeling paradigms should be used in various project contexts, in
particular, contract specifications, various types of executable models, for example,
finite-state machines, Kripke structures, etc. [32]. It is not obvious that the trans-
formation of models from one paradigm into another one will bring real benefit.
Each of the model kinds is suitable for analysis of specific aspects of the system
behavior, so we should not expect that, for example, a functional model will facili-
tate estimation of the execution time and memory required. However, obtaining
some skeleton or a prototype of the model of one kind on the basis of another kind
is quite possible.
The development of various tools for modeling and specification description for
the MBT purposes is required. In spite of the progress in the area of technologies
for development of Domain Specific Languages (DSL), practically, the systems
based on universal languages benefit from the large number of programmers know-
ing such languages. The same can be also said about monolingual systems – they
overtake multilingual ones.
Modern achievements in the area of static and hybrid static-dynamic analysis allow
integration of these techniques into the MBT systems, at that, the mod-
els/specifications as well as software implementations should be the subject of this
analysis.
580 A. K. Petrenko, V. Kuliamin and A. Maksimov
To overcome the problems with the extreme lack of specifications in real practice,
tools for work with requirements and models (see, for example, [33]), in particular,
with system models [34, 35] should be developed and deployed. For multicompo-
nent systems, MBT tools should be integrated with the tools for architecture and
process mining.
References
1. Bourdonov, I. B., Kossatchev, A. S., Kuliamin, V. V., Petrenko, A. K.: UniTesK Test
Suite Architecture. In: FME 2002. LNCS 2391, pp. 77–88. Springer-Verlag (2002)
2. Kuliamin, V. V., Petrenko, A. K., Kossatchev, A. S., Bourdonov, I. B.: The UniTesK Ap-
proach to Designing Test Suites. Programming and Computer Software, 29(6), 310–322
(2003)
3. Bourdonov, I. B, Kossatchev, A. S., Petrenko, A. K., Galter. D.: KVEST: Automated Gen-
eration of Test Suites from Formal Specifications. In: Proceedings of Formal Method Con-
gress, Toulouse, France, 1999. LNCS 1708, pp. 608-621 (1999)
4. Peleska, J.: Industrial-Strength Model-Based Testing – State of the Art and Current Chal-
lenges. Invited Talk. In: Petrenko, A. K., Schlingloff, H. (eds.) Proceedings Eighth Work-
shop on Model-Based Testing (MBT 2013), Rome, Italy, 17th March 2013. Electronic
Proceedings in Theoretical Computer Science, 111, pp. 3–28. DOI: 10.4204/EPTCS.111.1
(2013)
5. Börger, E., Stärk, R.: Abstract State Machines: A Method for High-Level System Design
and Analysis. Springer-Verlag (2003)
6. UniTESK technology, http://unitesk.ispras.ru
7. OLVER project, http://linuxtesting.org
8. Microsoft Interoperability Initiative, http://www.microsoft.com/openspecifications
9. Pakulin, N. V., Khoroshilov, A. V.: Development of Formal Models and Conformance
Testing for Systems with Asynchronous Interfaces and Telecommunications Protocols.
Programming and Computer Software, 33 (6), 316-335 (2007)
10. Khoroshilov, A. V.: Specification and Testing of Components with Asynchronous Inter-
faces. Candidate's thesis, Moscow (2006)
11. Kuliamin, V. V., Petrenko, A. K., Pakulin, N. V.: Extended Design-by-Contract Approach
to Specification and Conformance Testing of Distributed Software. In: Proceedings of
WMSCI'2005, Orlando, USA, July 10-13, 2005. Model Based Development and Testing,
v. VII, pp. 65-70 (2005)
12. Bourdonov, I. B., Demakov, A. V., Jarov, A. A., Kossatchev, A. S., Kuliamin, V. V.,
Petrenko, A. K., Zelenov, S. V.: Java Specification Extension for Automated Test Devel-
opment. In: Proceedings of PSI'01. LNCS 2244, pp. 301-307. Springer-Verlag (2001)
13. The Linux Foundation Consortium. LSB Certification Test Suite,
http://ispras.linuxbase.org/index.php/LSB_Certification_System
14. Maksimov, A. V.: Requirements-Based Conformance Testing of ARINC 653 Real-Time
Operating Systems. In: Proceedings of the Data Systems in Aerospace (DASIA 2010)
Conference, 2010. ESA SP-682, ISBN 978-92-9221-246-9 (2010)
15. Ivannikov, V. P., Kamkin, A. S., Kossatchev, A. S., Kuliamin, V. V., and Petrenko, A. K.:
The Use of Contract Specifications for Representing Requirements and for Functional
Testing of Hardware Models. Programming and Computer Software, 33(5), 272–282
(2007)
UniTESK: Component Model Based Testing 581
16. Chupilko, M. M.: Developing Test Systems of Multi-Modules Hardware Designs. ISSN
0361-7688, Programming and Computer Software, 38(1), 34–42, Pleiades Publishing,
Ltd. (2012)
17. Zelenov, S. V., Zelenova, S. A.: Model-Based Testing of Optimizing Compilers. In: Proc.
of the 19th IFIP TC6/WG6.1 International Conference on Testing of Software and Com-
municating Systems – 7th International Workshop on Formal Approaches to Testing of
Software (TestCom/FATES 2007). LNCS 4581, pp. 365–377. Springer-Verlag, Berlin
Heidelberg (2007)
18. Zelenov, S. V., Silakov, D. V., Petrenko, A. K., Conrad, M., Fey I.: Automatic Test Gen-
eration for Model-Based code Generators. In: IEEE ISoLA 2006 Second Intern. Sympo-
sium on Leveraging Applications of Formal Methods, Verification and Validation. Paphos,
Cyprus, pp. 68-75 (2006)
19. Kamkin, A. S.: A method of Automation of Simulation Testing of Microprocessors with
Conveyer Architecture Basing on Formal Specifications. Candidate's thesis, Moscow
(2009)
20. Kornykhin, E. V.: A Method of Automation of Testing Program Generation for MMU
Verification. Candidate's thesis, Moscow (2010)
21. Kamkin, A.S., Tatarnikov, A.: MicroTESK: An ADL-Based Reconfigurable Test Program
Generator for Microprocessors. In: Proceedings of the 6th Spring/Summer Young Re-
searchers’ Colloquium on Software Engineering (SYRCoSE 2012), May 30-31, 2012,
Perm, Russia (2012)
22. MBT Survey, http://www.robertvbinder.com/docs/arts/MBT-User-Survey.pdf
23. Grieskamp, W.: Microsoft’s Protocol Documentation Program: A Success Story for
Model-Based Testing. In: Testing – Practice and Research Techniques. LNCS 6303, p. 7
(2010)
24. Adams, C.: Safety-Critical Software for Mission-Critical Applications to Get Boost with
Release of DO-178C. Military & Aerospace Electronics, 10 (2010)
25. Common Criteria, http://www.commoncriteriaportal.org
26. SpecExplorer, http://research.microsoft.com/en-us/projects/specexplorer
27. The Java Modelling Language (JML), http://www.eecs.ucf.edu/~leavens/JML/index.shtml
28. Pakulin, N. V.: Integrated Modular Avionics: New Challenges for MBT. In: ETSI TTCN-3
User Conference and Model Based Testing Workshop, Bangalore, India, 11-14 June 2012
(2012)
29. Code Contracts, http://research.microsoft.com/en-us/projects/contracts
30. C++TESK, http://forge.ispras.ru/projects/cpptesk-toolkit
31. Kuliamin, V. V.: Component Architecture of Model-Based Testing Environment. Pro-
gramming and Computer Software, 36(5), 289–305 (2010)
32. Kuliamin, V. V.: Multi-paradigm Models as Source for Automated Test Construction. In:
Proceedings of the 1-st Workshop on Model Based Testing (MBT'2004, in ETAPS'2004),
Barcelona, Spain, March 27-38, 2004, Electronic Notes in Theoretical Computer Science,
111:137-160, Elseveir, (2005)
33. ReQuality tool, http://requality.org/en/doc.en.html
34. Khoroshilov, A. V., Albitskiy, D., Koverninskiy, I. V., Olshanskiy, M. Yu.,
Petrenko, A. K., Ugnenko, A. A.: AADL-Based Toolset for IMA System Design and Inte-
gration. SAE Int. J. Aerosp. 5(2) (2012)
35. Systems Modeling Language (SysML), http://www.sysml.org
Protoautomata as Models of Systems with Data
Accumulation
Irina Mikhailova1 and Boris Novikov2 and Grygoriy Zholtkevych2
1
Luhansk Taras Shevchenko National University,
Institute of Information Technology, 2, Oboronna Str., 91011, Luhansk, Ukraine
mia irina@rambler.ru
2
V.N. Karazin Kharkiv National University,
School of Mathematics and Mechanics, 4, Svobody Sqr., 61022, Kharkiv, Ukraine
{bvnovikov46,g.zholtkevych}@gmail.com
Abstract. In the paper formal models of software systems and their
components based on the notion of an abstract machine are discussed.
Necessity to model systems with data accumulation sets the problem of
study of generalizations of the notion of an abstract automaton. In the
paper two generalizations, namely, preautomata and protoautomata, are
considered. It is shown that passing from automata via preautomata to
protoautomata can be naturally realized using the language and methods
of category theory.
Keywords. system modelling, abstract automaton, category of automata,
preautomata, category of preautomata, globalization, protoautomaton,
category of protoautomaton, reflector, free protoautomaton
Key terms. MathematicalModel, SpecificationProcess, VerificationPro-
cess
1 Introduction
Theory of abstract state machines or abstract automata is widely applied in
different areas of Computer Science. While the early applications of automata
theory were connected with theory of compilators design (see, for example, [1]),
the more recent its applications are focused on the problems of specification and
verification of behaviour of software components [3, 11]. Such changing of the
object of the theory was marked by R. Milner in [11]: “In the classical theory,
rather little attention is paid to the way in which two automata may interact, in
the sense that an action by one entails a complementary action by another. This
kind of interaction requires us to look at automata in new light; in particular, this
interdependency of automata via their actions seems to demand a new approach
to behavioural equivalence”.
Protoautomata as Models of Systems with Data Accumulation 583
But the practice of modelling system behaviour based on the automata ap-
proach has shown that the approach is inadequate if data accumulation for the
correct response is necessary.
Using the concept of partial action of a semigroup on a set [7, 10], we have
defined the notion of preautomaton and studied its properties [4, 12]. The further
study has shown that preautomata can be used for modelling some aspects of
behaviour of systems with a delayed response [13, 14].
In this paper, we consider a more general class of automaton-liked systems —
the class of protoautomata. All necessary information from the theory of semi-
groups, automata theory, and category theory can be found in the monographs
[5, 6, 8, 9].
We use the notation ϕ : A 99K B for the partial mapping of A to B (unlike
the complete mapping A → B). If ϕ(a) is not defined for a ∈ A, we write
ϕ(a) = ∅. The free monoid on the alphabet Σ is denoted by Σ ∗ , and its unit by
ε. All actions and preactions used in the paper are right, as it is common in the
automata theory.
2 Preliminaries
We will use the definition of the automaton in the following form (the condition
of the finiteness is ignored):
Definition 1. Given a set X and a free monoid Σ ∗ over the alphabet Σ, an
automaton is a mapping X × Σ ∗ → X : (x, a) 7→ xa such that for all x ∈ X
and u, v ∈ Σ ∗
xε = x, (1)
x(uv) = (xu)v. (2)
More general concept is the following
Definition 2 (see [4]). A preautomaton is such a partial mapping of
X × Σ ∗ 99K X : (x, a) 7→ xa, that
a) the condition (1) is fulfilled;
b) if xu 6= ∅ and (xu)v 6= ∅, then x(uv) 6= ∅ and equality (2) is fulfilled;
c) if xu 6= ∅ and x(uv) 6= ∅, then (xu)v 6= ∅ and equality (2) is fulfilled.
The preautomata over the monoid Σ ∗ form a category PAut(Σ); its mor-
phisms are such maps ϕ : X → Y that
(∀ a ∈ Σ ∗ )(∀ x ∈ X)( xa 6= ∅ =⇒ ∅ 6= ϕ(x)a = ϕ(xa)). (3)
The category Aut(Σ) of the automata over Σ is a full subcategory of PAut(Σ).
Preautomata appear in the following situation. Let Y be an automaton and
X an arbitrary nonempty subset of Y . Then a restriction of an action on X is a
preautomaton.
Conversely, let X × M 99K X be a preautomaton. The construction which is
inverse to restriction is called globalization. More precisely:
584 I. Mikhailova, B. Novikov and G. Zholtkevych
Definition 3. A globalization of the preautomaton X is an automaton Z with
an injection ι : X → Z such that for all a ∈ Σ ∗ , x ∈ X
xa 6= ∅ =⇒ ∅ 6= ι(x)a = ι(xa),
ι(x)a ∈ ι(X) =⇒ xa 6= ∅ & ι(xa) = ι(x)a.
Obviously, ι is a morphism of PAut(M ). We also call it a globalization.
Definition 4. A globalization ι : X → Z is called universal if for any global-
ization ι0 : X → Z 0 there is an unique morphism κ : Z → Z 0 such that ι0 = κι.
The following construction gives an universal globalization (obviously unique up
to isomorphism) for any preautomaton X × Σ ∗ 99K X. Define a relation ` on
the set X × Σ ∗ :
(x, ab) ` (xa, b) ⇐⇒ xa 6= ∅. (4)
Let ' be an equivalence relation generated by `, and X U = (X × Σ ∗ )/ '. An
equivalence class of ' containing a pair (x, a) is denoted by [x, a]. For [x, a] ∈ X U
and b ∈ Σ ∗ , we set [x, a]b = [x, ab]. Thus a complete action on X U is defined.
Theorem 1. The automaton X U with a morphism ιU : X → X U : x 7→ [x, ε] is
the universal globalization of the preautomaton X.
Proof. See [4, Theorem 2] t
u
3 Protoautomata
The main object of this paper is a generalization of the notion of preautomaton:
Definition 5. A protoautomaton is a partial mapping X × Σ ∗ 99K X :
(x, a) 7→ xa such that
a) the condition (1) is fulfilled;
b) if xu 6= ∅ and (xu)v 6= ∅, then x(uv) 6= ∅ and equality (2) is fulfilled.
We will also denote the protoautomaton from this definition simply by X, if it
does not cause a confusion.
Example 1. Let S be a free subsemigroup of Σ ∗ and α : X × S → X an au-
tomaton. Define a partial mapping X × Σ ∗ 99K X as an extension of α, putting
xu = ∅ for u ∈ Σ ∗ \ S; so we get a protoautomaton over Σ ∗ . Note that in
general it is not a preautomaton. In addition, this example shows that the au-
tomaton over an infinite alphabet can be represented as a protoautomaton over
a two-letter alphabet.
Example 2. Let X = {x, y} be a two-element set, L a subset of Σ ∗ . Define a
protoautomaton X × Σ ∗ 99K X putting for a 6= ε
y, if a ∈ L,
xa =
∅, if a ∈
/ L,
and ya = ∅. This example shows that protoautomata recognize all languages.
Protoautomata as Models of Systems with Data Accumulation 585
We denote the category of protoautomata with morphisms defined by the
condition (3) by PtAut(Σ); clearly, PAut(Σ) is its subcategory.
It follows from the theory of partial action of semigroups [6, Theorem 5.7],
that a protoautomaton which is not a preautomaton has no globalization. More
precisely, for the protoautomaton X we can construct an automaton X U as in
Sec. 2, but in this case the morphism ιU is not injective in general.
In this situation, the concept of a reflector is useful. We recall its definition
[9]:
Definition 6. A subcategory D of a category C is called reflective if with each
object C ∈ C an object RD (C) ∈ D is associated (called D-reflector of the object
C) and a morphism ρD (C) : C → RD (C) (reflection morphism) such that for
each D ∈ D the diagram
ρD (C)
C −→ RD (C)
↓
D
can be extended uniquely to a commutative diagram by some morphism out
HomD (RD (C), D).
It is convenient to use another description of the equivalence ':
Lemma 1. Define a relation ] on the set X × Σ ∗ :
(x, a) ] (y, b) ⇐⇒ (∃ a0 , b0 , p ∈ Σ ∗ )(a = a0 p & b = b0 p & xa0 = yb0 6= ∅).
Let ≈ be the equivalence relation generated by ]. Then ≈ coincides with '.
Proof. If (x, a) ] (y, b) then
(x, a) = (x, a0 p) ` (xa0 , p) = (yb0 , p) a (y, b0 p) = (y, b),
whence ≈ ⊆ '.
Conversely, if (x, a) ` (y, b), then a = cb, y = xc for some c ∈ Σ ∗ . Hence
` ⊆ ]. Consequently, ≈ ⊇ ' t
u
Remark 1. Obviously, ` is reflexive and transitive, while ] is reflexive and sym-
metric.
Lemma 2. Let X be a protoautomaton, Y be a preautomaton (both over Σ ∗ ),
α : X → Y be a morphism, x, y ∈ X, a ∈ Σ ∗ . Then [x, ε] = [y, a] implies
α(x) = α(y)a 6= ∅..
Proof. It follows from the condition that
(x, ε) ] (z1 , b1 ) ] . . . ] (zn , bn ) ] (y, a)
for some z1 , . . . , zn ∈ X, b1 , . . . , bn ∈ Σ ∗ .
586 I. Mikhailova, B. Novikov and G. Zholtkevych
Apply induction on n. Since x = z1 b1 6= ∅ then α(x) = α(z1 )b1 . Suppose
that α(x) = α(zn )bn 6= ∅. By definition of the relation ]
bn = cp, a = dp, zn c = yd 6= ∅
for some c, d, p ∈ Σ ∗ . Then α(zn )c 6= ∅ and by the induction α(zn )(cp) 6= ∅.
Since Y is a preautomaton then
α(x) = α(zn )(cp) = α(zn c)p = α(yd)p = (α(y)d)p = α(y)a.
Proof has completed t
u
Similarly (and even easier) one can prove
Lemma 3. Let X be a protoautomaton, Y be an automaton (both over Σ ∗ ),
α : X → Y be a morphism, x, y ∈ X, a, b ∈ Σ ∗ . Then [x, a] = [y, b] implies
α(x)a = α(y)b.
Proof is omitted t
u
We set [X, ε] = {[x, ε] ∈ X U | x ∈ X}. Obviously, [X, ε], being a subset of
X U , is a preautomaton, and in addition, ιU (X) = [X, ε].
Theorem 2. Let X be a protoautomaton over Σ ∗ , then
1. [X, ε] is a reflector for X in the category PAut(Σ),
2. X U is a reflector for X in Aut(Σ),
3. X U is a reflector for [X, ε] in Aut(Σ).
Proof. 1) Let Y be some preautomaton and α : X → Y be a morphism of
protoautomata. The required morphism β : [X, ε] → Y is uniquely determined
from the equality α = βιU . Indeed, for x ∈ X we have α(x) = βιU (x) = β([x, ε]).
It follows from Lemma 2 that β is well-defined.
2) Similarly, using Lemma 3.
3) Follows from 1), 2), and the following well-known fact [9]:
If A ⊂ B ⊂ C are categories, A is reflective in B, and B is reflective in C, then A
is reflective in C. Moreover, the reflection morphism from C to A is the product
of the corresponding reflection morphisms from C to B and from B to A t
u
Corollary 1. Aut(Σ) is a reflective subcategory of PAut(Σ). Moreover, the
universal globalization of a preautomaton is its reflector.
Example 3. Let X = {x, y, z, t}, p, u, v ∈ Σ ∗ \ {ε}. We set zu = zv = t, z(up) =
x, z(vp) = y and sw = ∅ for all s ∈ X, w ∈ Σ ∗ \ {ε, p, u, v}. In such a
manner X turns into a protoautomaton. Since (x, ε) ] (z, up) ] (z, vp) ] (y, ε)
then [x, ε] = [y, ε] and the reflection morphism of X is non-injective.
A large class of protoautomata is contained in the following example.
Protoautomata as Models of Systems with Data Accumulation 587
Example 4. Consider a preautomaton X × Σ ∗ 99K X as a directed weighted
multigraph with states as vertices and with edges of the form (x, u, y), where
x, y ∈ X, u ∈ Σ ∗ , and y = xu. Let U be an arbitrary subset of edges of X. Build
a transitive closure U t of the set U , extending it step by step by the rule: if the
edges (x, u, y) and (y, v, z) are at some stage in the expansion, then on the next
step we include the edge (x, uv, z). Then U t is a protoautomaton.
Example 3 shows that there exists a protoautomaton such that it can not be
embedded into some preautomaton (and thus into some automaton).
4 Free Protoautomata
It is well known [2] that free automata play a significant role in the theory of
automata (for example, in the problem of constructing a minimal realization).
Therefore, it is advisable to consider the question about the existence of free
objects in the category of protoautomata.
Recall the necessary definitions:
Definition 7. Let C and D be categories, F : C D be a functor, C be an object
of C. An object D of D is called free on C with respect to the functor F ,
if there is a morphism α : C → F D such that for any object D0 ∈ D and any
morphism β : C → F D0 there exists the unique morphism γ : D → D0 such that
F (γ)α = β. (5)
We consider a category Rel(Σ) whose objects are pairs (X, ρ), where X is
a set (X ∈ Set), ρ ⊂ X × Σ ∗ is a binary relation such that X × {ε} ⊂ ρ.
A morphism φ : (X, ρ) → (Y, σ) of Rel(Σ) is a map φ : X → Y such that
(φx, u) ∈ σ for (x, u) ∈ ρ.
Next, let F be a forgetful functor F : PtAut(Σ) Rel(Σ) mapping each
protoautomaton X × Σ ∗ 99K X to the pair (X, ρ) with ρ = {(x, u) | xu 6= ∅}.
Theorem 3. For each object (X, ρ) ∈ Rel(Σ) there is a protoautomaton that is
free on it with respect to the forgetful functor F .
Proof. For (X, ρ) ∈ Rel(Σ) construct a protoautomaton M = (ρ × Σ ∗ 99K ρ),
defining the action by the rule
(x, uv), if (x, uv) ∈ ρ
(x, u)v =
∅, if (x, uv) ∈
/ ρ.
Then F M = (ρ, ρ̂), where ρ̂ = {((x, u), v) | (x, u)v = (x, uv)} ⊂ ρ × Σ ∗ . Define
the morphism α : (X, ρ) → (ρ, ρ̂) by the formula α(x) = (x, ε).
Let us show that M is a free protoautomaton on (X, ρ).
Let N = (Y × Σ ∗ 99K Y ) be some protoautomaton and F Y = (Y, σ). For the
required morphism γ : M → N of (5) we have:
γ(x, ε) = F (γ)(x, ε) = F (γ)α(x) = β(x).
588 I. Mikhailova, B. Novikov and G. Zholtkevych
Then for any u ∈ Σ ∗ one can obtain
γ(x, u) = γ(x, ε)u = β(x)u,
i.e. γ is uniquely determined t
u
5 Conclusion
It seems that the class of protoautomata, which has been introduced in the
paper, gives the most abstract models for systems with discrete behaviour. This
class of abstract machines includes not only machines reacting on the received
data immediately, as automata, but it also includes machines whose reactions
depend on the accumulated information.
The machines of this class having a greedy behaviour are united into a sub-
class whose instances are called preautomata. Machines of the subclass are used
for modelling behaviour systems for complex event processing as it was shown
earlier [13, 14]. This class of machines, in contrast to the class of automata, is
closed under structural decomposition, and hence, is more suitable for specify-
ing complex systems. But the condition c) in the definition of a preautomaton
(see Definition 2) seems unnatural. This condition also impedes definition of a
nondeterministic preautomaton.
Therefore, by eliminating the condition c) we provide a possibility to study
nondeterministic models. In our opinion, the models derived in this way (pro-
toautomata) are interesting objects that can be used for specification and veri-
fication of complex systems.
References
1. Aho, A.V., Ullman, J.D.: Theory of Parsing, Translation, and Compiling. Prentice-
Hall, New York (1972)
2. Arbib, M.A., Manes, E.G.: Machines in a category: an expository introduction.
SIAM Rev. 16, 163–192 (1974).
3. Börger, E., Stärk, R.: Abstract State Machines: A Method for High-Level System
Design and Analysis. Springer-Verlag, Berlin Heidelberg (2003)
4. Dokuchaev, M., Novikov, B., Zholtkevych, G.: Partial actions and automata. Alg.
and Discr. Math. Vol. 11, 2, 51–63 (2011)
5. Eilenberg, S.: Automata, Languages, and Machines, vol. B. Academic Press, New
York (1976)
6. Holcombe, W.M.L.: Algebraic Automata Theory. Cambridge Univ. Press (1982)
7. Hollings, C.: Partial actions of monoids. Semigroup Forum. 75, 293–316 (2007)
8. Lallement, G.: Semigroups and combinatorial applications. John Wiley, New York
(1979)
9. MacLane, S.: Categories for the Working Mathematician. Springer, Berlin (1971)
10. Megrelishvili, M., Schröder, L.: Globalization of confluent partial actions on topo-
logical and metric spaces. Topol. and Appl. 145, 119–145 (2004)
11. Milner, R.: Communicating and Mobile Systems: The Pi Calculus. Cambridge Uni-
versity Press, Cambridge (1999)
Protoautomata as Models of Systems with Data Accumulation 589
12. Novikov, B., Perepelytsya, I., Zholtkevych, G. Pre-automata as mathematical mod-
els of event flows recognisers. In: V. Ermolayev et al. (eds.) Proc. 7-th Int. Conf.
ICTERI 2011, 41–50 (2011)
13. Perepelytsya, I., Zholtkevych, G.: On some class of mathematical models for static
analysis of critical-mission asynchronous systems. Syst. ozbr. ta viysk. tehn. Vol.
27, 3, 60–63 (2011)
14. Perepelytsya, I., Zholtkevych, G.: Hierarchic Decomposition of Pre-machines as
Models of Software System Components. Syst. upravl. navig. i zv’iazku. Vol. 20,
4, 233–238 (2011)
Models of Class Specification Intersection of Object-
Oriented Programming
Dmitriy Buy1 and Serhiy Kompan1
Taras Shevchenko National University of Kyiv, Faculty of Cybernetics,
03680 Academician Glushkov Avenue 4d, Kyiv, Ukraine
buy@unicyb.kiev.ua, skompan@mail.ru
Abstract. This paper describes the application of heterogeneous algebraic sys-
tem for the construction of the formal model of object database instead of object
algebra. Complete formalization of the operation of intersection of class speci-
fications is given.
Keywords. object-oriented programming, object database, object algebra, class
specification
Key terms. MathematicalModel
1 Introduction
In applications of information technologies there is a problem of construction of the
so-called dependable and stable systems and infrastructures – the systems which be-
have stably under all, especially, critical working circumstances. Similarity of risks
and increasing actuality of their decline to an acceptable level for critical applications
led to the appearance of a special term “safeware”, by the analogy with the terms
“hardware”, “software”, “firmware” etc., which combines two components: safe –
secure and ware – a product, an item. This term was suggested and patented by the
leading expert of NASA on the questions of infrastructure security, professor
N. Leveson, who registered the appearance of a modern field of knowledge called
safeware engineering [1]. We mention a fundamental statement both obvious, and
elusive in its nature: it’s impossible to talk about stability of a working system, espe-
cially of the infrastructure, if there is no formal model of its operation which has been
constructed and verified. Moreover, for the construction of a formal model, more or
less complex, not “toylike”, there should exist a mathematical apparatus with the help
of which software developers create a formal model and verify it according to the
source demands of a customer could.
For the full confidence in the fact that informational system will work stably (will
be dependable and stable), one should single out system components, describe them
formally and verify. Indeed, nowadays there is nothing instead of a “divide and rule”
Models of Class Specification Intersection of Object-Oriented Programming 591
approach to cope with this difficulty. In fact, one of the most important components
of any complex system (infrastructure) is databases. That’s why there should exist an
appropriate formal model. For the relational databases such a formal model has been
already constructed and explored considerably. This issue is exhaustively covered in
the literature, beginning from the pioneering works by E.F. Codd (see, e.g. [2], the
first textbooks [3, 4] and modern textbooks [5, 6]). We mention only a collection of
works done by the collaborators of Taras Shevchenko National University of Kiev on
the natural generalization of classical results of the databases relational approaches [7-
14].
Nowadays, there are a lot of formal models of object-oriented databases (OODB)
[15-20]. Each of these models elaborates OODB to a certain extent by applying cer-
tain mathematical apparatus. The analysis of research papers dedicated to OODB has
shown that authors overlook the question arising from the necessity to construct a new
class specification with the two given specifications. For example, the construction of
a super class from two specified classes (the operation of intersection of class specifi-
cation), the construction of a subclass from two super classes (the operation of union
of class specification). The intersection of class specifications is important, in our
opinion, as it provides for the opportunity to construct the core of a new program with
two programs which allows integrating these two programs that results in the Frame-
work version. This paper is dedicated to the exploration of the operation intersection
of class specifications and refining conditions under which the intersection of classes
is possible.
2 Practical results
The authors of this paper have conducted a number of investigations in the field under
research: for example, in the article [21] it has been suggested to consider an object
algebraic system as a model. Formally it can be formulated like this:
, ; obj ; spec , , where is a set of objects’ classes, is a set of class speci-
fication, obj is a set of operations over objects, spec is a set of operations over
class specifications, and a relation is a partial order which formalizes in-
heritance. The main objective of this article is specification of the intersection opera-
tion and the difference of class specifications.
Let’s start with the intersection operation . Let us formalize the notion of a class:
by a class we mean a pair K s, , where s is a functional binary relation which
associates an attribute with its meaning (from a universal domain D ), and is a
functional binary relation, which brings to conformity a method with its signature.
Therefore [21], the relations s and determine a class specification.
The intersection operation (of class specifications) is an operation of the form
: , where: s1 , 1 s2 , 2 s1 s2 , 1 2 , where is a
standard set-theoretical intersection.
592 D. Buy and S. Kompan
We will demonstrate some results concerning the structure of a partially ordered
set (poset) F , , where is a set of all the functional binary relations (on the uni-
versal domain ), а is an ordinary set-theoretical inclusion. These results will sup-
plement the results of the paper [22]. All undetermined notions and designations are
understood in terms of this paper.
Lemma 1. For the arbitrary functional binary relations and the following equal-
ity is true: f g ( f g ) (domf domg) □
def
Proof. ■ Let us start with X domf domg . Let us use generally valid properties
of the set-theoretical restriction operation (a binary ratio on a set) (monotony, dis-
tributivity etc.) [19].
Firstly, we have an inclusion dom( f g ) domf domg X . Secondly, from this
the next chain of equalities and inequalities follows:
f g ( f g ) dom( f g ) ( f g ) X f X g X f g .
Thus, f g ( f g ) X ( f g ) (domf domg ) □
def
Below is a relation of consistency: f g f X g X , where
def
X domf domg . In [7] the main property of consistency was determined as:
f g f g is a functional binary relation.
The following lemma’s corollary forms another criterion of consistency.
Corollary (the criterion of consistency of functional binary relations). Let f , g be
def
arbitrary functional binary relations, and X domf domg . Then:
f g dom( f g ) X , ( f g ) dom( f g ) X . □
Proof. ■ The proof is performed by using a Lemma 1 and inclu-
sion dom( f g ) X . It’s important to note that the second (the first) equivalence is
a formal corollary of the first one (of the second one accordingly). □
As for the structure of the poset F , , there are two statements.
Statement 1. Poset F, is a lower semilattice, and at the same time,
inf f , g f g . □
The proof results from the fact that is a commutative idempotent semigroup and
from a well-known connection between such semigroups and lower semilattices (see,
e.g. [23]).
More complete information about the poset F , is given by the following
statement.
Statement 2. (the structure of poset F , ). The following statements are true:
1. The empty function f is the smallest element (“a bottom”)
Models of Class Specification Intersection of Object-Oriented Programming 593
2. The largest element in poset F , exists if and only if the universe D is single-
ton
3. The infimum exists for any nonempty set F and inf F f F f
4. The supremum of the set F exists if and only if in the case when the set F is re-
stricted, and sup F f F f
5. The element f is an atom only when f is singleton
6. Poset F , is a relatively complete poset and a complete (upper) semilattice □
Let’s proceed to the substantial interpretation of above results.
The operation constructs a new class which will be basic (paternal) for classes
arguments. This intersection can also be empty, in this case we will get a special
empty class.
As the relation on the specifications is component wise
( s, s, s s ) , all properties of the relation (statements 1, 2)
can be lifted to the relation . The corresponding formulations are obvious and
thereby are omitted.
3 Results and conclusions
The model of intersection operation of class specifications has been examined. This
operation has been specified as set-theoretical intersection. The specification f g
has been interpreted as the largest total part of f and g , that is, the specification
from which specifications-arguments can be obtained by inheritance (in other words,
the result specification is the specification of a paternal class). The conditions for
nonempty (equivalent, empty) intersection have been examined.
As for formal results, natural criteria of function consistency have been presented
(corollary) which supplement the already known criteria; the structure of a partially
ordered set of partial functions has been specified (statements 1, 2).
References
1. Kharchenko, V. S.: Safety of Critical Infrastructures: Mathematical and Engineering
Methods of Analysis and Ensuring. N.E. Zhukovsky National Aerospace University (2011)
(in Russian)
2. Codd, E. F.: A Relational Model of Data for Large Shared Data Banks. Comm. ACM, 13
(1970)
3. Maier, D.: The Theory of Relational Databases. Computer Science Press (1983)
4. Ullman, J., Garsia-Molina, H., Widom, J.: Database Systems: The Complete Book. Pren-
tice Hall Inc., Stanford (2002)
5. Kroenke, D. M.: Database Processing: Fundamentals, Design, and Implementation. Pren-
tice Hall (2011)
6. Date, C. J.: An Introduction to Database Systems. In: Addison-Wesley, (2000)
594 D. Buy and S. Kompan
7. Buy, D. B., Kahuta, N. D.: Full Image, Restriction, Projection, Relationship Compatibility.
Theoretical and Applied Aspects of Program Systems Development: International Confer-
ence, December 8-10, pp. 244-260 (2009) (in Ukrainian)
8. Buy, D. B., Bogatiryova, J. A.: The Theory of Multisets: Bibliography, Use the Table in
Databases. Radio Electronic and Computer Systems, 7(48), 56–62 (2010) (in Ukrainian)
9. Buy, D. B., Polyakov, S. A.: Compositional Semantics of Recursive Queries in SQL-like
Languages. Bulletin of Kyiv University. Series. Phis.-Math. Science, 1, 45–56 (2010) (in
Ukrainian)
10. Buy, D. B., Glushko, I. M.: Generalized Table Algebra, Generalized Tuple Calculus, Gen-
eralized Domain Calculus and Theirs Equivalence. In: Bulletin of Kyiv University. Series.
Phys.-Math. Science. 1, 86–95 (2011) (in Ukrainian)
11. Buy, D. B., Puzikova, А. V.: Completeness of Armstrong Axioms. In: Bulletin of Kyiv
University. Series. Phys.-Math. Science, 3, 103–108, (2011) (in Ukrainian)
12. Redko, V. N., Brona, J. Y., Buy, D. B., Polyakov, S. A.: Relational Databases: Tabular
Algebra and SQL-like Language. AcademPeriodika (2001) (in Ukrainian)
13. Buy, D., Silveystruk, L.: Formalization of Structural Constraints of Relationships in «En-
tity-Relationship» Model. In: Electronic Computers and Informatics 2006: International
Scientific Conference, September 20-22, pp. 96-101, Kosice, Slovakia (2006)
14. Buy, D., Glushko, I.: Equivalence of Table Algebras of Finite (Infinite) Tables and Corre-
sponding Relational Calculi. In: Proceedings of the Eleventh International Conference on
Informatics INFORMATICS’2011, November 16-18, pp. 56-60. Rožňava, Slovakia,
(2011)
15. Piskunov, А. G.: The Formalization of the Object-Oriented Programming Paradigm,
http://www.realcoding.net/dn/docs/machine.pdf (in Russian)
16. Piskunov, A. G.: The Formalization of the OOP: Types, Sets, Classes,
http://agp1.hx0.ru/articles/typeSetsClasses.pdf (in Russian)
17. Chaplanova, Е. B.: Operating Specification of Object-Relational Data Model.
Radіoelektronіka, Informatika, Upravlіnnya, 12, 75–79 (2011) (in Russian)
18. Richta, K, Toth, D.: Formal Models of Object-Oriented Databases. In: Objekty 2008. Žil-
ina: Žilinská univerzita v Žiline, Fakulta Riadenia a Informatiky, pp. 204-217,
http://www.ksi.mff.cuni.cz/~richta/publications/richta-toth-Objekty2008.pdf (2008)
19. Sarkar, M., Reiss, S.: A Data Model and a Query Language for Object-Oriented Database.
In: Island, Department of Computer Science Brown University Providence, Rhode, CS-92-
57, http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.34.4531&rep=rep1&type=
pdf (1992)
20. Gail, M., Shaw, S.: Zdonik A Query Algebra for Object-Oriented Databases. Island, De-
partment of Computer Science Brown University Providence, Rhode, CS-89-19
http://trac.common-lisp.net/elephant/raw-attachment/wiki/RelationalAlgebra/shaw89
query.2.pdf (1989)
21. Buy, D. B., Kompan, S. V.: Union and Intersection Operations of Classes Specifications in
Heterogen Algebraic System for Object-Oriented Programming. In: Proc. SWorld. Int. Sci-
Pract. Conf. Modern Problems and Solutions in Science, Transportation, Manufacturing
and Education. KUPRIENKO, Odessa, vol. 4, pp. 45–49 (2012) (in Russian)
22. Buy, D. B., Kahuta, N. D.: Properties Related Confinality and Order a Set of Partial Func-
tions. Bulletin of Kyiv University. Series. Phys.-Math. Science, 2, 125–135, (2006) (in
Ukrainian)
23. Skornyakov, L. A.: Elements of the Theory of Structures. Nauka, Мoskow (1982) (in Rus-
sian)
Author Index
A
E
Alferov, Eugene................................... 108
Alobaidi, Mizal...................................... 18 Echahed, Rachid ..................................521
Arkatov, Denis B. ................................ 178 Ermolayev, Vadim ...... II, 64, 99, 108, 295
Aronov, Andrey ................................... 252
G
B Glazunova, Olena G.............................411
Baiev, Oleksandr ................................. 118 Glukhovtsova, Kateryna ........................48
Baklanova, Nadezhda .......................... 550 Guba, Anton.........................................490
Batyiv, Andriy ....................................... 18
Becker, Karsten ................................... 424 I
Beletsky, Alexsander ........................... 352
Isomöttönen, Ville ...............................221
Beletsky, Anatoly ........................ 311, 352
Itkonen, Jonne......................................221
Beletsky, Evgeny ................................. 311
Iurtyn, Ivan ..........................................187
Bilousova, Lyudmyla........................... 209
Ivanov, Ievgen .....................................448
Blinov, Igor Ol..................................... 565
Bodnenko, Dmitry ............................... 281
Bonda, Darya....................................... 360 K
Buy, Dmitriy........................................ 590 Kandyba, Roman .................................352
Keberle, Natalya G.................................79
C Kharchenko, Vyacheslav .....................146
Klionov, Dmitriy M. ............................464
Chaabani, Mohamed ............................ 521
Kobets, Vitaliy ........................ II, 310, 329
Cochez, Michael .................................. 221
Kolgatin, Oleksandr .............................209
Kolgatina, Larisa..................................209
D Kompan, Serhiy ...................................590
Davidovsky, Maxim ...................... 99, 295 Kotkova, Vera......................................236
Derevianko, Andrii ................................ 30 Kravtsov, Hennadiy ................ II, 236, 410
Didenko, Ievgen................................... 118 Kropotov, Aleksandr..............................30
Doroshenko, Anatoliy............................ 38 Kryukov, Sergey ..................................310
Dzyubenko, Artem............................... 252 Kryvolap, Andrii..................................533
Kukharenko, Vladimir .................273, 410
596 Author Index
Kuliamin, Victor.................................. 573 Popov, Peter.........................................146
Kushnir, Nataliya................................. 195 Pratt, Gary L. ...........................................3
Kuzminska, Olena ............................... 264 Protsenko, Galina.................................264
L R
Lavrischeva, Ekaterina ........................ 252 Ralo, Aleksandr .....................................30
Lazareva, Elena ................................... 339 Richter, Harald.....................................424
Lazurik, Valentine ............................... 118 Romenska, Yuliia.................................130
Letichevsky, Alexander ........................... 4 Rudenko, Margarita .............................401
M S
Maksimov, Andrey .............................. 573 Schreiner, Wolfgang ............................533
Mallet, Frédéric ................... 130, 289, 475 Selyutin, Victor....................................401
Mantula, Elena....................................... 91 Semenyuk, Andriy ...............................393
Manzhulam Anna ................................ 195 Shushpanov, Constantin.......................490
Mashtalir, Vladimir ............................... 91 Shyshkina, Mariya ...............................436
Matthes, Ralph..................................... 506 Sitzmann, Daniel..................................424
Matzke, Wolf-Ekkehard .......................... 2 Sokol, Vladyslav....................................48
Mayr, Heinrich C.................................... II Spivakovska, Evgeniya ........................236
Mazol, Sergey...................................... 360 Spivakovskiy, Aleksander............... II, 236
Mazol, Sergey...................................... 366 Strecker, Martin ................... 447, 521, 550
Mesropyan, Karine .............................. 385 Styervoyedov, Sergiy.............................30
Mikhailova, Irina ................................. 582
Moiseeva, Oksana................................ 366 T
Möller, Dietmar P.F............................. 424
Morze, Natalia ..................................... 264 Tatarintseva, Olga ..................................64
Morze, Natalia V. ................................ 411 Tirronen, Ville .....................................221
Tkachuk, Nikolay...................................48
Tolok, Vyacheslav .................................99
N
Nikitchenko, Mykola .............. II, 447, 533 V
Novikov, Boris .................................... 582
Valko, Nataliya ....................................195
Varava, Anastasiia ...............................163
O
Vasylevych, Leonid .............................187
Odarushchenko, Oleg .......................... 146 Vozniy, Oleksiy .....................................30
Odarushchenko, Valentina................... 146
W
P
Weissblut, Alexander J. .......................374
Payentko, Tanya .................................. 310 Winckel, Mathias .................................506
Peschanenko, Vladimir ........... II, 447, 490
Petrenko, Alexander K......................... 573 Y
Petukhova, Lyubov .............................. 236
Yatsenko, Olena.....................................38
Models of Class Specification Intersection of Object-Oriented Programming 597
Z Zhereb, Kostiantyn.................................38
Zholtkevych, Galyna............................475
Zaporozhchenko, Yulia........................ 410 Zholtkevych, Grygoriy..... II, 18, 163, 475,
Zaretska, Iryna ..................................... 475 582
Zavileysky, Mikhail................................ II