A significant problem of modern society is preservation and increase of resources of human brains. A person must analyze huge amounts of data, in condition of existence of multiple obvious and hidden relations in incoming data and external influence factors in order to make correct decision. Nowadays, by the most part, data to be analyzed are results of measurements of technical and natural objects parameters. Modern artificial intelligence based information systems are not able to process efficiently enough such type of data. It requires develop a new class of adaptive and intelligent information systems that possess fundamentally new capabilities and provide new features. In the article general requirements to such systems are formulated, a conceptual model of their operation is suggested.
Nowadays significant attention of government is paid to the problem of preservation
and increase of resources of human brains in conditions of the increasing
environmental and information pressures and high responsibility for made decisions [
A relatively new challenge of data processing is lack of knowledge about problems to
be solved. Often experts in subject domains are not able to identify all points that
have to be checked and analyzed in the data and information streams.
An essential part of analyzed data are measurements of technical and natural objects
parameters. As shown in the report of IDC [
A new class of adaptive intelligent IS (AIIS) is expected to provide support for end users that have to deal with measurement processing and analyses tasks in the contexts of solving applied problems.
AIIS Tasks. To justify expectations an AIIS has to solve following tasks: i) reduce
amount of data due to its transformation to information or knowledge; ii) build linked
data and information space (ISp) and permanently support its actual state; iii) enrich
data, information and knowledge using all available sources; iv) provide machine
based applied problems solutions regarding measurements processing [
AIIS Properties. An AIIS must possess following properties: i) accumulating – to be able to support continuous systematic actualization of data, information and knowledge about the target and related subject domains; extend ISp by gathering all objective and subjective data, information and knowledge which can be correlated to the problems to be solved; ii) resource saving - to be directed to saving human brain resources, spend to understanding, analysis and estimation huge amounts of data, information and knowledge gathered by IS; iii) accessible – technologies must have low cost of ownership; iv) have theoretical background i. e. use classic mathematical theories, machine learning methods, methods of intelligent analysis and pattern recognition.
AIIS Distinguishing Features. AIIS uses a new approach to data processing and analysis, based on a new way of application of intelligent technologies.
AIIS is to be: i) intelligent – knowledge is used in all steps and phases of problems solving; ii) automatic - IT specialists are not necessary for solving data processing and analysis problems; iii) dynamic – data processing and analysis is realized in run time or real time mode, processing procedures can be adapted in accordance with changing context; iv) able to process historical data – to be able to mine useful information and knowledge from historical data taking into account historical context.
The IS with the described properties define a new class of IS, which can fit processes of measurements analyses according to the observed contexts. Intelligence of the systems allows solve applied end users tasks meaningfully using knowledge.
The modern tendencies of IS design, development and support are based on the
ideas of wide usage of models –architectural, information, technological and others.
The ideas have taken the form of a model driven (MDE) approach [
Development of the conceptual model has 3 goals: i) the conceptual model determines the scope of AIIS and can be conceded as a description of the AIIS class; ii) the conceptual model is to be used as a meta model for generating model of concrete AIIS, which can be used as a part of their architectural description; iii) the conceptual model is to be used as a knowledge base for storing AIIS community knowledge about models.
The proposed conceptual modal is based on a number of general ideas: 1. Real world objects are too complex for modeling, but their numerous views have simple models. 2. Real world processes are poorly predictable, too complex to be formalized, but well decomposable. 3. Feasible way to investigate the real world lies through dealing with measurements – gathering, storing, processing, analyzing. 4. Capability of consuming measurements is reachable if based on measurements progressive transformations.
Model principals. The following principles are to be used for model design. 1. The main value is knowledge; it is vital to operate with knowledge in each case. 2. Any data can be meaningful; thus, all data is supposed to be carefully processed. 3. Data that is out of date is harmful; actual state of data must be supported. 4. The resources of human minds have limits; data processing and analyzes must be organized in order to provide understanding of measurements streams. 5. Both models and processes must have internal connections and external links. 6. Models and processes must be understandable by humans and machines; they must have descriptions in terms of applied subject domains and must support standards. 7. Models and processes must be easily configured; configuration must be supported by different external tools, including GUI tools. 8. Models and processes must be hierarchical and consist of multiple simple elements. 9. Models and processes must be totally oriented on support of data and information transformations. 10. Models and processes must be adaptable at the level of structure and the level of contents. 11. Models and methods used for measurements processing must be highly flexible; they have to fit exactly spontaneously changing contexts without any delay. 12. Models and processes must correspond to the desires of consumers and available resources. 13. Models and processes must be open to integrate any external data, information, knowledge and processes.
Model foundation. Foundation for the model implementation form three technological
stacks: transformation technologies, semantic web stack and IT technologies for IS
design and support.
1. Transformations technologies. Transformation technologies relate to the domain of
data, in particular, measurements processing and analyses. They allow consider
processes of dealing with measurements as a sequence of transformations.
Transformations are defined for JDL models adapted for measurements processing (MJDL)
[
Semantic web technologies. Semantic technologies must be considered as a world
level agreement that is supported by multiple standards [
For AIIS an adaptive approach for systems architecture design within the agile
concept has been developed [
The conceptual model (Fig. 1) gives the general view on how measurements can be consumed by humans and machines. The list of the principle entities of the model, their types and references to the groups are given in the table 1, in the table 2 types of relations between entities are briefly described. Three groups are defined according to the entities material features: i) real entities (RE) - group of real world entities (objects of the real world, AIIS, tools and means); ii) model entities (ME) - group of creates used for data transmission used for data transmission used by AIIS for filling up IS
used by AIIS for filling up IS
defines
TMIS describes
M TMIS build based on
GM TMIS transfer Fig 1. The structure of the proposed model for measurements processing and analyses
GM MPA build based on
M MPA is build according to
M describes
M M Is used for filling up build based on used for processing and analysis
processes agrees with fills
GAM AIIS
GOM AIIS Is a part of
Is a part of
HAD AIIS Is build on the base of
AD AIIS describes
AIIS describe
fills
M AIIS is a part of SM AIIS is a part of
DM AIIS GM M used by AIIS for filling up IS MMeeaassuurreemmeennttssPPrroocceessssiinnggaannddaannaallyyssiiss AAIIIISSaarrcciitteeccttuurree
GGeenneerraallAArrcchhtteeccttuurraallssoolluuttiioonnssaannddmmeeaannssooffiittssIIMMpplleemmeennttaattiioonn
Is build by means of
AF AIIS
HAF AIIS build based on designs configures defines
End user fills
Sponsor
TToooollss designs
Architect -models (general models, AIIS models); iii) logical entities (LE) - logical domains (environment, domain of measurements processing and analyses).
Table 1. List of the principle entities of the conceptual model Entity id Entity name
GM MPA TS M TS GM TS TMIS M TMIS GM TMIS M M M GM M AIIS M AIIS SM AIIS DM AIIS AD AIIS HAD AIIS GAM AIIS GOM AIIS AF AIIS HAF AIIS E M E GM E M MPA target system model of the target system general model of the target system TMI system TMIE RE model of the TMI system ME general model of the TMI system measurements IPME RE model of measurements ME general model of the measurements adaptive intelligent information system SAE RE model of AIIS ME static model of AIIS dynamic model of AIIS architectural description of AIIS hierarchy of the architectural descriptions of AIIS general architectural models of AIIS general ontological models of AIIS architectural framework of AIIS AFE RE hierarchy of architectural frameworks of AIIS ME environment TSEE LE model of the environment ME general model of the environment model for measurements processing and analyses of TS MPAE ME parameters general model for measurements processing and analyses of TS model of the subject domain of measurements processing LE and analyses The types of the entities are set according to the real world objects that they relate to: i) target system entities (TSE) - entities related to the target system and its description; ii) TMI system entities (TMIE) - entities related to the TMI system and its description; iii) initial & processed measurements entities (IPME) entities related to initial and / or processed measurements and their descriptions; iv) system architecture entities (SAE) - entities related to the AIIS and its architecture; v) architectural framework entities (AFE) - entities related to architectural frameworks; vi) target system environment entities (TSEE) - entities related to target system environment; vii) measurements processing and analyses entities (MPAE) - entities related to measurements processing and analyses.
M GM MPA M MPA M MPA M E M
AIIS proand
TMI S MSD MPA M M M M
model entities of real objects are used for filling up AIIS models of information space TMI systems transfer measurements model of the subject domain of measurements processing and analyses defines general model for measurements processing and analyses of TS model of measurements is build according to model for measurements processing and analyses of TS model of measurements must agree with model for measurements processing and analyses of TS AIIS fills model of measurements; AIIS fills model of AIIS AIIS processes measurements model for measurements processing and analyses of
TS is used in AIIS for dealing with measurements is build according to agree with fills processes used for cessing analyzes to be part of is built by mean of
ME M AIIS dynamic model of AIIS and static model of AIIS part
of AIIS model AD AF AIIS architectural description of AIIS is build used archi
AIIS tectural framework of AIIS Description of the conceptual model. A Target system (TS) is a system to be observed or investigated. Environment (E) is a context for TS, it defines conditions in which TS operates. A TS is used only for technical systems. For investigation of natural systems E concept is applied. A technical system is described by a target system model (M TS), for a natural system an environment model (M E) is build. A M TS is a set of models which describe TS. Models of end systems are built on the base of generalized models (GM TS and GM E).
Results of parameters measurements of all systems are transferred to AIIS by means of telemetric IS (TMIS). A model of a TMIS (M TMIS) is a set of models which describe the TMIS. M TMIS is based on a generalized model of telemetric systems (GM TMIS). Results of TS parameters measurements (M) are described in accordance with measurement models (M M), which are built on the base of generalized measurement model (GM M). Processing of results of measurements is executed in accordance with models for measurements processing and analyses of TS parameters (M MPA). Similar to models of TS and models of TMIS models for measurements processing and analyses are based on generalized models (GM MPA). The list of methods and algorithms to be used in AIIS is defined by a subject domain model of measurements processing and analyses (MSD MPA).
Architecture description of AIIS (AD AIIS) is made in accordance with standard [
Main AIIS stakeholders are end users, architects, sponsors and researchers. An end user is an expert of an applied subject domain that operates technical systems or is an expert who solves problems related to natural systems. An architect is an IT specialist or group of specialists who design AIIS with the use of AF. Sponsors fill the repository of the general models with new models or modify existing models. Researchers of natural systems and environment create new models / modify existing models that describe natural objects. Researchers develop new methods and algorithms. MKB is available for both researchers and developers. It is of primary importance for all architects and analyst. They can form their own models with the help of a toolbox. The toolbox is a GUI instrument with the help of which a user can populate MKB, choose necessary models from the list of available reference models, transform and aggregate them in order to receive necessary target model. Received models are used as an element of an architectural description of the target system. So, a user receives a validated model and minimizes risks of dealing with a wrong model and can save time for model development.
The generalized scenario of the AIIS operation is shown in Fig. 2. An AIIS gathers results of measurements and use them for building measurement model. Activities initiated by the users are event based. A user forms requests for measurements processing and analyses. According to the request, model of AIIS is build. The model defines the structure and the content of DIK which are required to be received about the target natural or technical object as a result of the carried experiment. Then the model of measurements that reflect the expected raw data and results of its processing and analyses is created. For building a measurement model the models of target system and TMI system are required.
AIIS operating mode TS
TMI S M AIIS structure
M M structure
M M M
M M M AIIS Id AV SV CV
ObjV DIKV ВV (PR) SvcV AV PV
Fig 2. Generalized scenario of the AIIS operation
Measurements processing is a linked sequence of transformations of raw data. As the transformations go, the measurement models and AIIS models are filled. Often for making transformations external information and knowledge about the target system is to be used. Description integrates all view points; defines architectural context considers a system as an aggregation of interacting subsystems; is used as a structural description of the system defines potential capabilities of a system for solving applied problems describes a system in terms of used models describes a system in terms of an object model considers basic structures for data, information and knowledge representation and a set of specialized representations dependent on the solved problems considers a system as a set of working scenarios, executable activities, supported business processes considers a system as a set of services, describes functionality of а system at different levels of abstraction considers architecture of a system; processes of system architecture design, systems development and support considers a system as a set of platforms used for its implementation considers actual technical standards, methodologies, instructions, restrictions with which the system complies Permanent activity User request
bility) Viewpoint Model Viewpoint Object Viewpoint Data, Information and Knowledge Viewpoint Behavior (Process) Viewpoint Services Viewpoint
6
The list of projections contains views of the concept model that provide its overall
detailing. They are coordinated with the standard [
Levels of JDL model
Infrastructure services
System services
Buisness services Level 0 Signal/ Feature Assessment
Level 1
Entity Assesement
Level 2 Situation Assessment
Level 3
Impact Assessement
Level 4
Process Assessement
Level 5 Cognition Refinement Stakeholders Sponsor
End User Architect As an example it is suggested to consider a service-oriented projection (Fig. 3). This model is most closely related to the implementation tasks. SvcV describes AIIS as set of services and business processes (BP). For the services 4 levels of functioning are defined. Infrastructure services ( f ) are usually elementary services. They are fragments of program code, i.e. java code. Each infrastructure service is associated with the certain level of JDL model. System services ( y) are composite services. These services implement BP that assumes calling both Infrastructure and other System services. A System service as a rule is connected with 2 neighboring level of JDL model. Business services (b) are composite services which are described in terms of subject domain. They are to be formed on the base of system and other business services. External (high-level) (x) services that are targeted to support B2B interactions be also defined. They are similar to business services but without GUI. There are 4 different types of services: functional services (s) , DIK services (d) , interpreters (engines) (e) and supporting services ( p) . Functional services implement target procedures of processing and analyses. DIK services provide access to DIK resources. Engines are interpreters of scripting languages. Supporting services are services that support life cycle, security, etc.
The services are distributed among levels of JDL model. Each business process (BP) in AIIS is considered as service, and each service can be realized as BP. In the context of SvcV business process is a sequence of calls of services described in BPEL style. BP can be either static or dynamic. The general idea of the proposed Service Process JDL (SPJDL) model is given in Fig.3. For Service description following notation can be suggested. A service is described as Tlevel jdl , where T - type of services, level - level of services, jdl - level of JDL model; T {s, d, e, p} , T {s, d, e, p} , level { f , y, b, x}, jdl [0; 5] .
Different stakeholders work on different levels. An End user has an access only to Business Services. An Architect by the most part works with system and business services. Sponsors usually have rights to receive information about services in MKB and add new services into MKB. Сommonly they work with Infrastructure services. Developers and analyst have full access to all levels. 7
In the paper an approach for solving a problem of analysis huge amounts of incoming data, in condition of existence of multiple obvious and hidden dependencies and external factors is discussed. The main idea of suggested approach to AIIS development is creation of AIIS conceptual model, which is to be used as meta model. Application of the conceptual model allows formalize procedures of creating models of applied systems of the domain level. Using transformations developers can form their own models by detailing projections of the conceptual model. The model saves time for model development by means of model reuse and allows receive reliable validated end models. They can be reused both in the frame of product lines and independent AIIS.
The discussed approach provides support for end users who have to deal with measurement processing and analyses tasks in the contexts of solving applied problems by means of reducing amount of data and increasing its utility. It is achieved due to building linked data and information space, using all available sources of information and knowledge and data enriching techniques, support machine learning procedures regarding measurements processing. Procedures for measurements processing and analyses can be adapted in accordance with operative and historical context. It simplifies the tasks of exerts of data processing and analysis domain at the stage of system design
The approach can be implemented using existing technologies in combination with semantic technologies.
For creation of the conceptual model our practical experience in realization projects in different subject domains such as monitoring state of complex technical systems, intelligent GIS systems, medicine was used. All these systems were based on jdl model, detailed analysis show that they have many common features and use similar models.
At present it is used in the project of medical data processing and analyses for
Federal Almazov North-West Medical Research Centre [
Future activity is planned to be directed to population the conceptual model with new concrete models and adding new transformation features.