A generalized framework for concordance/discordance-based multi-criteria classification methods

This paper reviews multiple criteria classification methods (or multi-criteria classifiers), particularly those based on concordance/discordance concepts. The concordance refers to an aggregated metric indicating the truthfulness of a proposition according to a coalition of criteria. The discordance is an aggregated metric representing the strength of the opposition coalition to the truthfulness of the proposition. A generalized framework is proposed to synthesize the underlying computation algorithms for each classifier. In this paper, we argue the benefits of cross-fertilization of multiple criteria classification methods and information fusion algorithms.