Determining the Trustworthiness of a Case-Based Reasoning Solution

Case-based reasoning (CBR) infers a solution to a new problem by searching a collection of previously-solved problems for cases which are similar to the new problem. The collection of past problems and their associated solutions represents the CBR system's realm of expertise. The trustworthiness of a CBR solution depends on whether the new problem falls within this realm. An indication of the trustworthiness of a CBR solution allows the user to decide whether to use the solution. This paper proposes a set of decision criteria which indicate trustworthiness, based on the relative similarities of the items in the case library