多源不确定信息的随机模拟聚合评价方法及应用  

Stochastic Simulation Integrated Method for Multi-source Uncertain Information and Its Application

作　　者：王露 易平涛[2] 李伟伟[2] Wang Lu;Yi Pingtao;Li Weiwei(School of Management,Shenyang University of Technology,Shenyang 110027,China;School of Business Administration,Northeastern University,Shenyang 110167,China)

机构地区：[1]沈阳工业大学管理学院,辽宁沈阳110027 [2]东北大学工商管理学院,辽宁沈阳110167

出　　处：《中国管理科学》2024年第5期103-112,共10页Chinese Journal of Management Science

基　　金：国家自然科学基金项目(72301062,72171040,72171041);中央高校基本科研业务费项目(N2006013,N2006007)。

摘　　要：针对综合评价中多源不确定信息共存的评价问题,提出一种融合集成框架,并讨论了该框架的随机聚合求解方法。首先,对多源不确定信息按信息类别进行分类整合,构建多源不确定信息集成框架;然后,将各类别的多源不确定信息转化为统一范围内,按特定分布随机生成数据,并计算其对原始多源不确定信息的隶属程度。通过充分的模拟仿真,对信息集成框架进行求解;最后,基于回归树的方法得到被评价对象的可能性排序。本研究通过构建信息集成框架的方式对多源不确定信息,尤其是只包含部分被评价对象的片段信息、残缺信息的融合问题进行了探索,拓展了综合评价的实际应用范围。Under the increasingly complex and uncertain evaluation environment,the expression of uncertain information has been further developed and many uncertain theories and methodologies have been introduced into comprehensive evaluation problems.A comprehensive evaluation problem involves such issues that the experts with different knowledge backgrounds usually need the freedom to provide the opinions by their individual preferences and the development of social platform and search software has made it possible to obtain diversified information,such as the text evaluation information and fragment information that people leave on the website inadvertently.Besides,the absolute ranking,which means that an alternative is superior to its next adjacent one with 100%probability,is lack of explanation for comprehensive evaluation problems containing multi-source information especially uncertain information.It is a meaningful and urgent issue to propose a novel method to integrate multi-source inputs to a reasonable output.For the sake of the problems above,a multisource uncertain information(MSUI)integration framework was built to fuse all kinds of information mentioned in this paper by simulation techniques.Then,obtain the most likely ranking with pairwise priority probabilities.Specifically,the first problem is fusing the MSUI.The MSUI is normalized into a unified scope where the random numbers considering membership degree are generated by a certain distribution.The second problem is integrating the MSUI.The MSUI is classified by the types into different clusters,based on which the MSUI integration framework is established.The third problem is obtaining a reasonable output.the priority comparison of each alternative can be done by each simulation.After adequate simulation,the ratio of the times that one alternative is prior to another one to the total simulation times tends to be stable.Then,the pairwise priority matrix(PPM)is obtained.Based on the method of regression tree,the most likely ranking with pairwise priority pr

关 键 词：综合评价 多源不确定信息 信息集成框架 随机模拟 可能性排序 

分 类 号：C934[经济管理—管理学]