The outranking approach in Multiple Criteria Decision Analysis (MCDA) uses ranking procedures to exploit a fuzzy outranking relation, which captures the decision maker's notion of a ranking. However, as decision problems become more complex and computer performance improves, new ranking procedures are needed to rank complex data sets that decision-makers may not interpret. This paper discusses recent efforts and potential directions for developing ranking procedures that use multiobjective evolutionary algorithms (MOEAs) to exploit a fuzzy outranking relation. After that, based on the cooperative coevolutionary algorithms (CCEA) approach, we suggest some fundamental modifications to extend the RP2-NSGA-II+H algorithm that improve the scalability of this MOEA to exploit large-sized fuzzy outranking relations. Empirical results indicate that adjustments improve the RP2-NSGA-II+H algorithm for the addressed problem. The proposed ranking procedure outperforms RP2-NSGA-II+H in terms of ranking error rates based on the experiments conducted. Our experimental results also demonstrate that the proposed approach can be scaled for instances of the ranking problem of up to one thousand alternatives.

中文翻译：

---

一种利用大型模糊排名关系的协同进化算法


多标准决策分析 （MCDA） 中的排名胜出方法使用排名过程来利用模糊排名关系，从而捕获决策者的排名概念。然而，随着决策问题变得更加复杂和计算机性能的提高，需要新的排名程序来对决策者可能无法解释的复杂数据集进行排名。本文讨论了开发使用多目标进化算法 （MOEA） 来利用模糊排名关系的排名程序的最新努力和潜在方向。之后，基于协作协同进化算法 （CCEA） 方法，我们建议对 RP2-NSGA-II+H 算法进行一些基本修改，以提高该 MOEA 的可扩展性，以利用大规模模糊排名关系。实证结果表明，调整改进了针对所解决问题的 RP2-NSGA-II+H 算法。根据所进行的实验，所提出的排名程序在排名错误率方面优于 RP2-NSGA-II+H。我们的实验结果还表明，所提出的方法可以针对多达一千个备选方案的排名问题进行扩展。