Formalization of the KADS Interpretation Models

Abstract

For the KADS knowledge acquisition method, interpretation models are intended to guide and validate the transformation of verbal data into the conceptual model. Existing suggestions of interpretation models are limited in their usefulness due to lack of completeness, clarity, and precision. We present an attempt to denote interpretation models from KADS publications in a formal language, which we have formerly developed for notation of conceptual models. Formal notation guarantees an unambiguous meaning and enhances the usefulness of interpretation models for further formalization steps. We treat two distinct aspects of the problem, which may be understood as related to the expert’s and the knowledge engineer’s knowledge perspective: The first aspect is to provide a very concise and general notation to be readable for the expert. The second is to provide expansions of the terms in the concise language in variants of 1st order predicate logic (1stPL). Two important insights in the process of formalizing a considerable number of interpretation models were that one highly expressive intermediate sort model between the inference layer of KADS and the variety of possible domains could be created and that the number of knowledge sources seemed to converge when the number of interpretation models was getting larger.