共识驱动的概率语群言分类决策模型  

Consensus-driven Sorting Model for Group Decision-making with Probabilistic Linguistic Information

作　　者：朱峰丽 张世涛[1] 刘小弟[1] ZHU Feng-li;ZHANG Shi-tao;LIU Xiao-di(School of Microelectronics and Data Science,Anhui University of Technology,Ma'anshan 243032,China)

机构地区：[1]安徽工业大学微电子与数据科学学院,安徽马鞍山243032

出　　处：《模糊系统与数学》2024年第4期92-107,共16页Fuzzy Systems and Mathematics

基　　金：国家自然科学基金项目(72074001);国家社会科学基金项目(22BGL211);教育部人文社会科学青年基金项目(18YJC630249);安徽省自然科学基金项目(2108085MG240)。

摘　　要：分类群决策指多个决策者依据方案评价值,将有限个待评价的方案分配到预先定义的类的过程。在这个过程中,决策者评价的复杂不确定表征以及群体分类观点的冲突协调是其挑战。鉴于此,本文针对概率语言信息下的多属性分类群决策问题,考虑到决策者分类阅值的逆向优化,提出一种共识驱动的分类群决策模型。首先,综合考虑个体决策者分类向量和分类结果中每一类所含方案数的差异,定义新的序分类相似度。其次,基于提出的序分类相似度,构建两阶段分类共识模型,即第一阶段确定与决策者分类相似度最大的初始群分类结果;第二阶段考虑到初始群分类结果可能不唯一,从偏好和阅值两个共识角度建立共识驱动的分类群决策模型。最后,用一个算例说明分类决策模型的应用,灵敏度分析和比较分析验证模型的有效性和合理性。Classification-based group decision-making problem refers to the process in which decision makers assign a finite set of alternatives to predefined and ordered classes based on the assessment of alternatives.The challenges in the process are the complex uncertain representation of decision maker's evaluations and the coordination of conflicting group sorting opinions.Hence,aiming at classification-basedmulti-attribute group decision-making problem with probabilistic linguistic information,in this paper a consensus-driven sorting model considering the inverse optimization of the decision maker's sorting thresholdis proposed.Firstly,a novel sorting similarity is defined by taking into account individual decision maker's sorting vector and differences in the number of alternatives contained in each class of sorting results synthetically.Secondly,based on the proposed ordinal classification similarity,a two-stage sorting consensus model is constructed.In the first stage,the initial group sorting result with the maximum sorting similarity to the decision makersis determined.In the second stage,if the initial group classification results are not unique,a consensus-driven group decision model is established from the perspectives of preference and threshold.Finally,an example is given to illustrate the application of the sorting decision model.Sensitivity analysis and comparative analysis verify the validity and rationality of the model.

关 键 词：分类群决策 概率语言术语集 序分类相似度 共识 分类阅值 

分 类 号：C934[经济管理—管理学]