Cross-entropy measure for precedent case selection in case-based reasoning (CBR)

Abstract

In case based reasoning the most important phase is related to the use of precedent cases. In some domains of inquiry such as the legal domain, the selection of a precedent case assumes critical importance. Researchers in case.,based reasoning (eRR) appear for have not paid sufficient attention to the process of precedent case selection from among a large set, perhaps due to the fact that humans are poor all it as subjectiveness come in. This paper proposes a selection criterion based on cross-entropy, a fundamental information theoretic measure and is almost directly applicable to the problem of case selection in CBR, Besides, It is a general measure applicable to a large class of domains.