Scholarlink

<p>The multicriteria ranking problem involves ranking a set of alternatives in decreasing order of preference. The ELECTRE III method for solving this problem consists of two phases. The first phase constructs a fuzzy outranking relation known as the credibility matrix. The second phase exploits this matrix through the distillation procedure to derive a final ranking. The purpose of this paper is to explain why the distillation procedure needs to be revised and, secondly, to present the revised version we propose. This new version considers a different type of preference information from the decision-maker and modifies specific computational rules of the former method. The revised procedure replaces qualification scores with preference counts, i.e., the number of alternatives strictly preferred over each alternative, thereby improving stability and interpretability. This change improves the method's ability to handle ties, reduces sensitivity to minor inconsistencies, and ensures more faithful representations of decision-maker preferences. Furthermore, the procedure addresses the issue of cyclic preferences by systematically breaking cycles, contributing to more consistent rankings. The revised procedure is particularly valuable in complex multicriteria decision-making scenarios where decision-makers require robust, consistent, and interpretable rankings.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Determining the final recommendation in the ELECTRE III method with a revised distillation procedure

  • 期刊论文

  • 发表日期：

    2026年5月2日

查看全文
  • Juan Carlos Leyva López

摘要

The multicriteria ranking problem involves ranking a set of alternatives in decreasing order of preference. The ELECTRE III method for solving this problem consists of two phases. The first phase constructs a fuzzy outranking relation known as the credibility matrix. The second phase exploits this matrix through the distillation procedure to derive a final ranking. The purpose of this paper is to explain why the distillation procedure needs to be revised and, secondly, to present the revised version we propose. This new version considers a different type of preference information from the decision-maker and modifies specific computational rules of the former method. The revised procedure replaces qualification scores with preference counts, i.e., the number of alternatives strictly preferred over each alternative, thereby improving stability and interpretability. This change improves the method's ability to handle ties, reduces sensitivity to minor inconsistencies, and ensures more faithful representations of decision-maker preferences. Furthermore, the procedure addresses the issue of cyclic preferences by systematically breaking cycles, contributing to more consistent rankings. The revised procedure is particularly valuable in complex multicriteria decision-making scenarios where decision-makers require robust, consistent, and interpretable rankings.