Learning through estimating optimal formats for problem solving modules

An entire case retention in case-based reasoning is not necessarily capable of realizing what modules lead to problem-solving mal-performance. Moreover, learning through case retention calls for a subsequent case induction, which in turn may lead to a high computational cost. To circumvent this problem a new approach to learning is proposed that concentrates on estimating the optimal formats for CBR modules before getting into the main problem-solving process. In this respect, our objective is to estimate the optimal formats for case representation, case retrieval, and solution adaptation in order to upgrade problem-solving performance for future problems. Within this context, we will demonstrate that the learning phase in CBR can itself be performed using another process of CBR.