=Paper=
{{Paper
|id=Vol-125/paper-19
|storemode=property
|title=Achieving Enterprise Model Interoperability through the Model-Based Architecture Framework for Enterprises
|pdfUrl=https://ceur-ws.org/Vol-125/paper16.pdf
|volume=Vol-125
|authors=H.D. Jørgensen,O.P. Ohren
}}
==Achieving Enterprise Model Interoperability through the Model-Based Architecture Framework for Enterprises==
https://ceur-ws.org/Vol-125/paper16.pdf
Achieving Enterprise Model Interoperability through the
Model-Based Architecture Framework for Enterprises
Håvard D. Jørgensen, Oddrun P. Ohren
SINTEF ICT, PO Box 124, Blindern, N-0314 Oslo, Norway
{hdj,odo}@sintef.no
Abstract. This paper describes an ontology for enterprise modelling, The
ontology has enabled conceptual integration of two different modelling
methodologies, one based on UEML (Unified Enterprise Modelling Language),
the other a UML profile for enterprise modelling for software development. The
ontology, called Model-based Architecture Framework for Enterprises
(MAFE), extends a general framework for model based architectures (MAF). It
contains a minimal number of core concepts, which were found to be
designated by numerous different terms in the two pre-existing methodologies.
1 Background and Motivation - The MAF Framework
MAF [1] is a framework for describing enterprise and ICT architectures, based on the
IEEE 1471 standard [2]. IEEE defines 'architecture' as the collection of the essential
components of the enterprise, their interrelations and evolution over time [2]. MAF
frameworks consist of:
• a conceptual model (ontology) of architectural descriptions depicting the core
concepts to be used for describing architectures;
• a set of structuring rules, specifying which views and models should be included in
an architecture description;
• a terminology covering the subject domain;
• a set of language mappings between the ontology and modelling languages
(terminologies) to be used to create architecture descriptions;
• a methodology, specifying how to go about developing architecture descriptions;
• principles of conformance, consistency, specialisation and realisation;
• a framework for an experience well repository for reusing knowledge.
While the concepts in the ontology should be defined independently of the viewpoints
of particular stakeholders, terminologies and modelling languages may reflect the
concerns and viewpoints of particular groups of stakeholders.
An ontology is a specification of a conceptualisation. Concepts are units of
thought. Concepts can have a linguistic representation as terms in a language, and
refer to objects and properties in the physical world. The meaning of a concept is
determined by its reference set (the physical and ideational objects it refers to), and its
intension. The intension specifies the properties that apply to the concept [3]. Two
concepts are synonyms if both their reference set and their intension are equal.
Concepts can be categorised according to logical strength (ability to express facts
precisely) into indefinite individuals, definite individuals, classes, relations and
quantitative properties. Over the lifecycle of a model, all of these concept categories
may appear, typically evolving from vague concepts towards increased logical
strength.
Conceptual models must be able to cater for all of these degrees of vagueness and
precision. Perspective integration [4] requires that different degrees of specificity
may be expressed with the same concepts. We thus choose to include in the
conceptual model of MAFE only the core concepts, and allow local terminologies to
represent these concepts in different ways, at different degrees of specificity and
logical strength. The core principle is that concepts directly refer to ideas or physical
things, properties and phenomenon, while terms capture the current state of a
description of these concepts. Well-formedness rules define what is a correct model,
not what may occur in the world, and should thus be defined for terms, rather than
concepts.
2 Conceptual Models of Enterprises
MAFE is a refinement of MAF. Whereas MAF covers any kind of system, MAFE
focuses on the specific kind of systems called ‘enterprise’, defined as a project,
undertaking or business activity. The conceptual model of MAFE is a refines that of
MAF. All MAFE models contain terms that designate MAFE concepts, which are
depicted in Figure 1.
• Enterprise is an undertaking. It is the system of analysis for an enterprise model. It
contains other enterprise elements.
• Task is a unit of work at any level of granularity or specificity. Terms like action,
process, activity, work item etc. typically designate different kinds of tasks. Such
term reflect different points of view, and are thus terms, not concepts.
• Resource is anything needed to perform work. Resource is actually a point of view
on objects that primarily exist as something else, e.g. persons, documents, and
things. Resources are specialised depending on their role
− Actor, the active subjects that perform the work, (humans or computers).
− Information, objects that bring knowledge to the process.
− Tool, objects applied by the actor to perform the work.
− Object, material things that are manipulated or applied in the work.
In addition to these resources, Knowledge, Time and Money are also important
for an enterprise. These aspects are intertwined with the resources, e.g. determining
availability.
• Decision, a choice among alternatives. A decision may also include e.g. the
identification and exploration of alternatives, and the timing of selections.
• Dependency is the general kind of association between enterprise elements.
Among tasks, dependencies capture sequencing and coordination needs, while
dependencies between tasks and resources entail that the resources are applied for
performing the task. Two important dependencies are defined. Fills relates an
element and a resource, signifying that one takes the role that the other specifies,
while Flow signifies that something should occur before the other something else.
• Goal (from MAF), representing visions, objectives, goals of actors, tasks,
enterprises etc.
SystemElement System
(from M AF) (from IEEE1471)
Goal
(from M AF)
+DependsOn Enterprise
has
*
Enterprise Money
Element
*
Knowledge
Person
Dependency Task Decision Resource
Fills Flow Actor Information Object Tool
Figure 1. MAFE enterprise ontology.
All of these objects may be decomposed into objects of any kind. For instance, a
dependency may be decomposed into a process of tasks that manage this dependency,
and a resource may be decomposed into a decision tree that articulates which actors
may fill the role.
2.1 Application of MAFE to UEML and UML
MEML is a modelling notation for the concepts described above. MEML has roots in
data flow diagrams, which was extended with resource modelling and control flow
semantics in APM [5]. Later the language was refined for the EXTERNAL model
driven infrastructure (as EEML [6]) and integrated with other languages in the UEML
(Unified EML) project [7]. Finally, we simplified and cleansed the language in the
Monesa project [8] (MEML = Monesa EML).
Like MAFE is a specialisation of MAF for enterprise architectures, so is MAFIS
for information systems. The environment viewpoint in MAFIS describes what
surrounds the information system, including a number of enterprise elements. MAFIS
was developed for a particular purpose in a government agency. We were able to
define all the enterprise terms in MAFIS with MAFE concepts. An example (from
organisation modelling) is shown in the table below (with MAF and MAFE concepts
in italic). By defining MAFIS terms (model element classes) in this way, we were
thus able to align the UML profile with a standard enterprise modelling language. In a
similar manner, we were also able to show that the enterprise viewpoint terminology
of ISO RM-ODP (Reference model for open distributed processing [9]) also
designates MAFE concepts.
MAFIS Term MAFE Concept
Organisation unit Composite Actor
Parent_organisation Decomposition of Actor
Structural relationship Dependency between Actors
3 Directions for Further Work
We are now completing the operationalisation of the MAFE methodology. The
metametamodel structure and services is of particular concern in this work, since it
controls how users may define new concepts, terms, modelling elements and views
needed in their local domain. In general, we are concerned about how modelling can
be made more user-friendly and less computer-oriented, allowing different degrees of
formalisation. Towards this aim, we have found reflective instances (containing their
own definition) to be a promising approach, removing much of the complexity of
class-oriented approaches. To make the framework comprehensible, we seek to
minimise the number of meta-layers, offering the users templates that they may copy
rather than classes to instantiate. Finally, computer-oriented encapsulation, where the
behavioural semantics of each class is defined inside the class and hidden from the
rest of the model, is replaced by semantic holism, where the meaning of each model
element may depend on the whole rest of the model. This approach better mimics the
flexibility, contextuality, simplicity and expressive efficiency of natural language [8].
References
1. Aagedal, J. Ø., et al., Model-Based Architecture Framework, v. 3.0. 2003, SINTEF,
Norway.
2. IEEE, Standard 1471-2000 - IEEE Recommended Practice for Architectural Description of
Software-Intensive Systems. 2000.
3. Bunge, M., Philosophy of Science - Vol.1 From Problem to Theory. Revised ed. Science and
Technology Studies, ed. M. Bunge. 1998, New Brunswick, USA: Transaction Publishers.
4. Opdahl, A.L. and G. Sindre, Facet Modelling: An Approach to Flexible and Integrated
Conceptual Modelling. Information Systems, 1997. 22(5): p. 291-323.
5. Carlsen, S., Conceptual Modeling and Composition of Flexible Workflow Models, PhD
thesis 1997, Norwegian University of Science and Technology: Trondheim, Norway.
6. Krogstie, J., Jørgensen, H.D., Interactive models for supporting networked organisations,
CAiSE 2004.
7. Zelm, M. Towards User Oriented Enterprise Modelling - Comparison of Modelling
Languages. in Concurrent Engineering (CE) Conference. 2003. Madeira, Portugal: A.A.
Balkema Publishers, Rotterdam, Netherlands.
8. Jørgensen, H.D., Interactive Process Models, PhD thesis 2004, Norwegian University of
Science and Technology: Trondheim, Norway.
9. ISO, Information Technology - Open distributed Processing - Reference Model - Enterprise
Language. 2002, ISO/IEC 15414: Geneva, Switzerland.