11
Features of formalization of experts’ judgments about criterion indicators of the project requests in evaluation decision making on the base of introformation models on fuzzy sets
Authors: E.M. Medvedieva, A.V. Evdokimova

Number of views: 409
Problem. Today there is a number of unsolved tasks for expert systems on of projects and project requests evaluation. Basic from them are tasks of determination the reliability and advantages of different methods to form a knowledge base, to select experts, to extract in short time enough knowledge from experts who separated geographically, to examine in short time plenty of project requests.
Purpose. To find features of experts’ judgments formalization about criterion indicators of the project request implementing introformation models on fuzzy sets for evaluation.
Design/methodology/approach. Theoretical research of connections and laws of the project request evaluation using methods of computer experiment, minimax ranging, multicriterion scales and main points of the non-force interaction theory in fuzzy raising.
Findings. Approach is worked out to form a knowledge base of the expert system on the project request evaluation in purpose to make a decision about their sponsorship.
The mathematical tool of the non-force interaction theory allows calculating the final entire mark of the project request taking into account all indicators on all determined amount of criteria. The offered model of such evaluation first envisages two-stage application of two groups of introformation indicators of the non-force interaction theory in the context of evaluation task: base and conditional entire marks. The offered system of multidimensional indicators of these marks allows simplifying procedure of programmatic realization of the proposed approach to formalization the experts’ knowledge.
The base absolute entire mark of the project request is characterized by the certain degree of membership to the fuzzy number 0,5. Such degree of membership is possible to determine according to ranging results of formed portfolios.
The decline of human factor influence at ranging is arrived due to application of minimax ranging method. Method most adequately implements the integral operating of entire in natural thinking processes of a person.
In the subsystem of automatic knowledge extraction in expert system of the project requests evaluation the tool of experts’ knowledge formalization should match the requirements of entire operating and of simple personal judgments in simple situations forming. From these positions it’s possible to formulate requirements to window interfaces referring: project portfolio ranging, forming the indicators system on corresponding criterion, forming of scales for quantitative and quality indicators of the project request. Successful application of the offered approach to formalization of experts’ knowledge can be attained, if system of forming the project requests database will be built on analogical principles. The construction of such system needs additional researches.
Originality/value. Offered approach to automated forming of knowledge base belongs to productional approach that is based on determination of rules "if… is equal to..., then mark is …". Due to this the approach is maximally close to integral person’s perception of any object without using (in the formalized explicated kind) of comparison procedures for evaluation.