A Hybrid KRS to Treat Fuzzy and Taxonomic Knowledge

摘要

The purpose of this paper is to present a hybrid Knowledge Representation System (KRS) in which Terminological Logic (TL) and Fuzzy Logic (FL) resources are used to store and to retrieve information. Knowledge here must be related to technical subjects that deals with terms whose meanings are vague and whose definitions are dependent on taxo-nomic organization of other terms, such as demographic census, medical diagnosis etc. Terminological and Assertional Knowledge compose the Knowledge Base (KB). The Terminological Knowledge defines crisp and fuzzy terms by means of TL term constructors. The Assertional Knowledge describes the world by means of Predicate Calculus formulae whose variables are annotated by TL expressions. The inference engine is able to answer questions that include Natural Language (NL) fuzzy quantifiers such as several, some, most, many etc. The advantages to be gained by this hybrid approach are: the ease of expressing knowledge and of retrieving information where the definition of fuzzy terms depend on several factors (for example, the definition of the fuzzy term tall for human beings depends on the height, the sex and the age of individuals); the contribution of using subsumption to improve the information retrieval process in goals that are structured in terms of NL fuzzy quantifiers.