ADAPTATION KNOWLEDGE ACQUISITION IN A CBR SYSTEM

Abstract

In a case-based reasoning system, adaptation is a complicated task since it requires domain-specific knowledge, which is generally difficult to define. To acquire such knowledge, we propose a semi-automatic approach based on Formal Concept Analysis (FCA) techniques. We use Logical Concept Analysis (LCA), a generalization of FCA, to extract adaptation conditions that enhance the retrieval and adaptation processes. In this paper, we present this approach, that has been implemented in COBRA, our ontology-based CBR platform, and applied to the diagnosis of gas sensor failures.