Adaptation cost as a criterion for solution evaluation

Abstract

In most case-based systems, the best case is selected by calculating similarities between the surface features of the cases. Sometimes, the most similar case is not the one that is the easiest to adapt. On the other hand, judging the adaptation effort can be tedious. In this paper, we propose case-adaptability to be used as an evaluation criterion as opposed to using it in the case-retrieval phase. Our method can be seen to combine the computational effectiveness of similarity-based retrieval with the better quality solutions of adaptation-guided retrieval.