Incremental concept learning and case-based reasoning: For a co-operative approach

Abstract

In case-based reasoning systems, the hierarchical memory organization presents the same set of problems as conceptual clustering learnt hierarchies. The co-operation between these two artificial intelligence methodologies gives several advantages. In case-based reasoning, memory and reasoning are closely linked. So the quality of the reasoning process is highly dependent upon that of the memory. Incremental concept learning permits to optimize the memory structures and organization. It also enlarges the application range of case-based reasoning. In incremental concept learning, case-based reasoning proposes its whole architecture, and above all its memory, from which the learning process can benefit, particularly for remedying the problem of the dependence upon the order of presentation of the instances, which is a crucial problem for this type of learning. It also permits the use of theoretical knowledge to explain the concepts learnt. This article presents a case-based reasoning system taking advantage from this co-operative approach.