Scholarlink

<p>With the increasing prevalence of online services, large-scale group decision-making (LSGDM) has become common. However, achieving consensus remains a significant challenge due to the diversity of opinions. This paper proposes a two-stage dynamic consensus model to facilitate agreement in multi-attribute LSGDM while addressing non-supportive behaviors. Using a novel maximum clique-based clustering method, the approach first groups decision-makers based on opinion similarity and then evaluates consensus within and among these subgroups. If the consensus level is insufficient, a dual-path dynamic feedback mechanism is applied, based on the urgency of the decision, to adjust opinions while considering trust relationships. The model dynamically updates trust and clustering throughout the process. Additionally, the consensus threshold can be adjusted during the consensus-reaching process to prevent potential errors. The proposed method is implemented in R and demonstrated through a case study, validating its effectiveness and applicability.</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

A two-stage bidirectional consensus model for large-scale group decision-making with maximum clique-based clustering

  • 期刊论文

  • 发表日期：

    2025年9月8日

查看全文
  • Azadeh Zahedi Khameneh,
  • Mehrdad Ghaznavi

摘要

With the increasing prevalence of online services, large-scale group decision-making (LSGDM) has become common. However, achieving consensus remains a significant challenge due to the diversity of opinions. This paper proposes a two-stage dynamic consensus model to facilitate agreement in multi-attribute LSGDM while addressing non-supportive behaviors. Using a novel maximum clique-based clustering method, the approach first groups decision-makers based on opinion similarity and then evaluates consensus within and among these subgroups. If the consensus level is insufficient, a dual-path dynamic feedback mechanism is applied, based on the urgency of the decision, to adjust opinions while considering trust relationships. The model dynamically updates trust and clustering throughout the process. Additionally, the consensus threshold can be adjusted during the consensus-reaching process to prevent potential errors. The proposed method is implemented in R and demonstrated through a case study, validating its effectiveness and applicability.