Expert Opinions and Expert Systems

by Felix S. Wong, Weidlinger Associates, Palo Alto, CA, USA,
Weimin Dong, Weidlinger Associates, Palo Alto, CA, USA,
Auguste Boissonade, Weidlinger Associates, Palo Alto, CA, USA,
Timothy J. Ross, Weidlinger Associates, Palo Alto, CA, USA,



Document Type: Proceeding Paper

Part of: Electronic Computation

Abstract:

The inference mechanism in most current expert (rule-based production) systems is based on exact matching of the antecedent or consequent of a rule with a fact or hypothesis. The rule will not be triggered when there is the slightest discrepancy in matching. However, expert opinions are not always crisp as can be attested by commonly used terms such as large, slight, etc. in the rules. Furthermore, the same term such as large may not have the same meaning when used by different experts. Such fuzziness in expert opinions and rules appears inconsistent with a crisp exact-matching inference mechanism. The paper describes the use of fuzzy sets to model vague and imprecise opinions, the formulation of rules from fuzzy opinions, and an inference mechanism for fuzzy rules. Within this fuzzy set representation framework, matching no longer needs to be exact. Partial matching is possible, and conflicting rules can be accommodated.

Subject Headings: Fuzzy logic | Expert systems | Mathematics | Mathematical models | Artificial intelligence (AI)

Services: Buy this book/Buy this article

 

Return to search