Representing domain structure of many-sorted Prolog knowledge bases

Abstract

After a brief introduction on the necessity of an explicit domain description for logic knowledge bases and the advantages of many-sorted logics, we argue that domain representation may consist of a separate logic theory which allows sorts to be assigned and tested dynamically. Then we show how this theory may be used by a meta-interpreter to implement many-sorted unification. Moreover we introduce a way for structuring the domain of discourse in a semantic network, leading to a system where domain knowledge and object language assertions are conceptually distinguished but embedded within an homogeneous formalism, which we call DRL (Declarative Representation Language). We call the former terminologic knowledge and the latter assertional knowledge, showing how some of the ideas of knowledge representation systems like KRYPTON can be successfully introduced within a logic programming approach.