Process modeling in conceptual categories

In this paper, we describe an approach to knowledge representation as an active process. More than merely recording knowledge in some extractable form, we recognize that representation must include the processes that perform the activities of insertion as well as extraction. The main result is a representational method that is powerful enough and flexible enough that systems built using it can represent, reason about, and change their own structure and behavior. The approach is based on recent work on knowledge representation for our "Wrapping" integration infrastructures, and on some comparative semiotics: studies of mathematical and linguistic reasoning.