基于证据理论的部分属性偏好信息群体赋权方法

发布时间：2018-11-25 08:30

【摘要】：针对已有的赋权方法在处理部分属性偏好信息的群体赋权问题上的局限性,提出了基于证据推理建模的解决方法。将不同专家的偏好信息视作来自不同信息源的证据,首先定义了焦元识别方法,然后采用惟一参照物比较判断法计算证据的基本可信度分配,应用Dempster规则进行证据合成,进而得到专家群体的属性赋权。在证据合成过程中,推导出了合成结果的解析表达式,与直接运用证据合成规则相比计算量大大缩小;在专家群体信息集结中,给出了一种基于位置权向量和证据相似度指数的客观证据赋权方法,从而突出了与大多数专家意见相似的专家的作用。最后给出了一个示例。
[Abstract]:In view of the limitations of the existing weighting methods in dealing with the problem of group weighting of partial attribute preference information, a method of evidence-based reasoning modeling is proposed. The preference information of different experts is regarded as evidence from different information sources. Firstly, the method of focal element identification is defined, then the basic reliability distribution of evidence is calculated by using the unique reference comparison method, and the evidence is synthesized by Dempster rule. Then the attribute weight of expert group is obtained. In the process of evidence synthesis, the analytical expression of the synthesis result is deduced, and the calculation amount is greatly reduced compared with the direct application of the rule of evidence synthesis. In this paper, an objective evidence weighting method based on position weight vector and evidence similarity index is presented, which highlights the role of experts who are similar to most experts. Finally, an example is given.
【作者单位】：东北大学工商管理学院;
【基金】：辽宁省财政科研基金资助项目(2007B002)
【分类号】：C934

【参考文献】