Retrieval in CBR Using a Combination of Similarity and Association Knowledge

Abstract

Retrieval is often considered the most important phase in Case-Based Reasoning (CBR), since it lays the foundation for the overall performance of CBR systems. In CBR, a typical retrieval strategy is realized through similarity knowledge and is called similarity-based retrieval (SBR). In this paper, we propose and validate that association analysis techniques can be used to enhance SBR. We propose a new retrieval strategy USIMSCAR that achieves the retrieval process in CBR by integrating similarity and association knowledge. We evaluate USIMSCAR, in comparison with SBR, using the Yahoo! Webscope Movie dataset. Through our evaluation, we show that USIMSCAR is an effective retrieval strategy for CBR that strengthens SBR.