A multiobjective evolutionary algorithm for deriving a final ranking from a medium-sized fuzzy outranking relation

The exploitation stage of a fuzzy outranking relation of the outranking methods aims to deal with the intransitivities in such a relation to consolidate and synthesise the decision maker's preferences between pairs of actions to obtain a ranking that is representative of his/her preferences over a given set of actions. In this paper, we propose this exploitation phase as a multiobjective optimisation problem and use a modified version of the multiobjective genetic algorithm (we refer to it as RP2-MOGA+H) to exploit a medium-sized fuzzy outranking relation to determining a partial pre-order of alternatives. To measure the performance of RP2-MOGA+H, we present an empirical study over a set of simulated ranking problems, which shows that RP2-MOGA+H can effectively exploit a medium-sized fuzzy outranking relation. Moreover, the RP2-MOGA+H outperforms other ranking procedures based on multiobjective evolutionary algorithms in the conducted experiments.