基于证据理论的部分属性偏好信息群体赋权方法  

A Method of Determining Weight with Partial Preference Information of Attributes in Group Evaluation Based on Dempster-Shafer Theory

作　　者：姚爽[1] 郭亚军[1] 黄玮强[1] 

机构地区：[1]东北大学工商管理学院,沈阳110004

出　　处：《系统管理学报》2010年第3期241-247,271,共8页Journal of Systems & Management

基　　金：辽宁省财政科研基金资助项目(2007B002)

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。Aiming at the limitation of existing weight determining methods on dealing with partial preference information of attributes in group evaluation,this paper proposes a method based on Dempster-Shafer（D-S） theory.Taking the preference information from different experts as proofs from corresponding sources,firstly,the definition is given to distinguish the core.Then,the preference information of different experts are transferred to the basic belief/probability assignment of proofs and the Dempster rule is used to combine these proofs,so group compromise attributes weight is calculated.The analysis formula of the combined results is deduced,so the calculation work is simplified.During the integration of group preference information,an objective weight determining method of proofs is given based on position weighting and the similarity indices of proofs.So it gives prominence to the experts whose opinions are similar to most of others.Finally,an example is illustrated.

关 键 词：群体评价 群体赋权 部分偏好信息 证据理论 

分 类 号：C934[经济管理—管理学]