Bipolar ranking from pairwise fuzzy outrankings

Abstract

In this paper we propose to apply the concept of ℒ-valued kernels
(see Bisdorff & Roubens[3, 4]) to the problem of constructing a global ranking from a pairwise ℒ-valued outranking relation defined on set of decision alternatives as encountered in the fuzzy preference modelling context (see Roy & Bouyssou [11] for instance). Our approach is based on a repetitive selection of best and worst candidates from sharpest ℒ-valued or most credible initial and terminal kernels (see Bisdorff [6]). A practical illustration will concern the global ranking of movies from individual evaluations by a given set of movie critics.