Generalizing Term Subsumption Languages to Fuzzy Logic

During the past decade, knowledge representation research in AI has generated a class of languages called term subsumption languages (TSL), which is a knowledge representation formalism with a well-defined logic-based semantics- Due to its formal semantics, a term subsumption system can automatically infer the subsumption relationships between concepts defined in the system. However, these systems are very limited in handling vague concepts in the knowledge base. In contrast, fuzzy logic directly deals with the notion of vagueness and imprecision using fuzzy predicates, fuzzy quantifiers, linguistic variables, and other constructs. Hence, fuzzy logic offers an appealing foundation for generalizing the semantics of term subsumption languages. Based on a test score semantics in fuzzy logic, this paper first generalizes the semantics of term subsumption languages. Then, we discuss impacts of such a generalization to the reasoning capabilities of term subsumption systems. The gener-alized knowledge representation framework not only alleviates the difficulty of conventional AI knowledge representation schemes in handling imprecise and vague information, but also extends the application of fuzzy logic to complex intelligent systems that need to perform high-level analyses using conceptual abstractions.