Weighted fuzzy production rules

Most of the research works on fuzzy production rules (FPRs) show that its knowledge representation power suffers from a serious shortcoming in that all propositions in the antecedent part are assumed to have equal importance (the lack of ‘local weight’ concept), and that a number of rules executed in an inference path leading to a specified goal or the same rule employed in various inference paths leading to distinct final goals may have relative degrees of importance, has not been explored (the lack of ‘global weight’ concept). This paper presents a weighted FPR (WFPR) incorporating the concepts of local and global weights. An improved method to compute the fuzzy value and the certainty factor of the consequent assertion and a better way to interpret the linguistic meaning of the consequent are proposed for this WFPR. Our approach offers the advantages of enhancing the knowledge representation power of an FPR, reducing the undesirable effects when computing the certainty factor of the consequent part by generalizing the traditional method of computation, and overcoming the tedious steps involved in Zadeh's compositional rule of inference (CRI) method. As far as interpretation of the linguistic meaning of the consequent part is concerned, our method is found to be better than the ad hoc approach of CRI-based methods.