MATRIX REPRESENTATION OF FUZZY PREDICATES AND ITS APPLICATION IN EXPERT-SYSTEMS

An approach to the construction of diagnostic expert systems which relies on a special matrix representation of fuzzy predicates in an attribute model of the subject area is proposed in this paper. The sectionalized matrix is the fuzzy analog of the conjunctive normal form of predicate representation. To form a knowledge base, one can use rules, examples, and counterexamples. By ''diagnostics problem'' one refers to the obtaining of information about the possible values of some attributes when data about the values of other attributes are available. The fuzzy resolution operation which is used for finding simple disjuncts serves as the basis for logic inference. The EDIP expert system-hull is realized on the basis of the presented formalisms.