Weighted fuzzy similarity relations in case-based reasoning: A case study in classification

This paper describes a fuzzy similarity relation approach to Case-Based Reasoning. Residuated implication operators are used to create a fuzzy resemblance relation between cases, modeling the CBR basic principle “the more similar the problem descriptions are, the more similar the solution descriptions are” as a fuzzy gradual rule. We take the classification of Schistosomiasis prevalence estimation in a region of Brazil as case study, in order to investigate the effects in such a framework of weighting cases individually in classification tasks, considering a set of training strategies.