Systems using Case-based Reasoning: Indexation strategy

Case-Based Reasoning (CBR) uses similar past situations to solve current problems. A major difficulty of CBR systems is getting them to choose automatically the references that will qualify each situation (case). Thus, the indexing process is primordial. The objective of this article is to clarify what is needed to choose an appropriate indexing strategy. We present the results of a comparative study of seven important indexing techniques used in CBR systems. Based on the characteristics identified in a recent survey of literature, we explain the reasons for using these techniques or developing new ones. Moreover, a strategy based on a combination of several techniques is a solution that aims to avoid their disadvantages and increase the indexing process performance. Finally, techniques allowing a memory representation of the case base, such as inductive ones, should be considered first.