EXPLOITING PRIOR KNOWLEDGE FOR EVALUATING CBR SYSTEMS USING A COVERAGE & CONSTRAINED EVIDENTIAL CLUSTERING BASED MODEL

Knowledge resource evaluation presents a concern of widespread interest in intelligent and knowledge management systems. For instance, the competence of Case Based Reasoning (CBR) systems in solving new problems depends mainly on the concept of cases coverage of the problem space. In this paper, we propose a new model for Case Base competence estimation that is based on cases coverage and partitioning, and enables the exploitation of available prior knowledge. This kind of background, which is handled in form of pairwise constraints, may be offered by domain-experts to aid the automated learning process. Actually, it is also important to manage the uncertainty involved by CBR systems' knowledge, since they reflect real-world situations. Our proposal tackles this latter problem under the belief function framework.