Knowledge Adaptation : a Means of Knowledge Acquisition

Abstract

Knowledge Adaptation is an alternative means of knowledge acquisition from previous experience in similar application domains. Adapting knowledge from an old expert system in some domain means transforming rules into rules useful for problem solving in a new, similar domain. The transformation follows from an analogical mapping of the situation described in a rule to situations in the new domain. This paper describes the Knowledge Adaptation approach to knowledge acquisition. We use diagnosis as the exemplar domain task, MOLTKE1 as the computational testbed, and CNC-machines2 as the exemplar application domain.