On the Automation of Case Base Development from Large Databases

Recent applications of Case-Based Reasoning (CBR) in industry have highlighted two major difficulties with developing CBR systems. These are, the integration of types of knowledge which are orthogonal to the knowledge represented by cases and the case engineering bottleneck. Exploiting the affinity between Data Mining and CBR we propose a unifying framework addressing these issues. Emphasising the automation of the entire CBR process we outline the Data Mining paradigms that may be employed and present initial results. Introduction Representing knowledge similar to that held by humans, within a computer system, has been one of the main foci in Artificial Intelligence research. Traditional rule-based approaches were found to have limitations especially in the areas of maintenance, brittleness and knowledge acquisition. CBR entered the Artificial Intelligence arena offering solutions to some key problems that had been plaguing the Rule-Based paradigm. The study of CBR is driven by two ..