DEQSION SUPPORT EXPERT SYSTEM DEVELOPMENT ON THE BASIS
OF THE FUNCTIONAL KNOWLEDGE-ORIENTED MODEL
KV Golovan, PhD., associated professor
National Aerospace University named after N.K Zhukovsky ”KhAl"
Nowadays there are some restrictions in the decision support expat system development technology. Such restrictions requirerepresenting knowledge in terms of one or several existing formalisms. However, such kind of knowledge representation makesa problem for decision support expert system developers and makes it necessary to involve knowledge engineers at the stage of domain knowledge formalization. Together with mentioned above in order to solve such kind of problems it is proposed to develop the models for knowledge representation, processing and acquisition based on the interaction between typical functional knowledge-oriented blocks.Thus, there were developed the functional blocks that are able to solve analytical and logical tasks as well as the tasks related to abstraction, specification, classification, finding of regularities and others. The development of such blocks is based on the KADS-methodology knowledge sources. The functional blocks have the configurable behavior algorithms and interact with each other during the process of finding solutions by the specific set of relations. There were also developed the mechanisms of interaction between the typical functional blocks with the traditional knowledge representation formalisms such as production systems and frames. Interaction with a frame model is performed by using the service inputs of functional blocks that make it possible to define the value of frame slots, rearrange the frame structure and the whole frame hierarchy, making it “living”. On the basis of such functional blocks the models of the problems difficult for formalization in different domains can be built, that makes it possible to increase the effectiveness of intelligent system development and reliability of the decisions made.The architecture of the instrumental framework consists of the next basic subsystems: conceptual modeling, visual modeling, algorithmic subsystem, design subsystem and administrative subsystem.The visual modeling subsystem makes possible to represent the processes of knowledge mining and processing as interaction of the typical functional blocks. Algorithmic subsystem provides the work of functional blocks and includes algorithms of classification, finding of regularities, solving of analytical problems, logic deduction and many others. The model design process is followed by the automatic generation of program code in the specially developed functional knowledge-oriented language.In such a way the framework provides the development of the decision support expert systems of different level and purpose.