Bipolar ranking from pairwise fuzzy outrankings

In this paper we propose to apply the concept of L-valued kernels to the problem of constructing a global ranking from a pairwise L-valued outranking relation defined on a set of decision alternatives as encountered in the fuzzy preference modelling context. Our approach is based on a repetitive selection of best and worst candidates from sharpest L-valued or most credible initial and terminal kernels. A practical illustration will concern the global ranking of movies from individual evaluations of a given set of movie critics.