Acquiring and managing knowledge using a conceptual structures approach: introduction and framework

One powerful representational system-conceptual structures that continue the trend towards knowledge programming-is examined. J.F. Sowa's conceptual structures approach (Conceptual Structures: Information Processing in Mind and Machine, Addison-Wesley, NY, 1984), a graph-based semantic network, is a general representational system with a natural-language expressivity. This allows problem-domain issues to be readily captured and expressed at several levels of granularity. Conceptual graph (CG) propositions can also act as a flexible intermediate form for the more common representations used in implementing many AI systems-frames, production rules, conceptual dependency, and logic systems. Thus, CG knowledge models are capable of symbolic translation into forms that are more familiar. Issues of expressively capturing the semantics of problem domains are discussed.>