一种改进的二分图投影算法及其在专家意见综合中的应用

Research on an Improved Bipartite Network Algorithm and Its Application in Synthesis of Expert Opinions

复杂问题的解决,单靠数据、信息和模型是不够的,还需集成专家的意见,然而专家的意见未必一致,需要对专家意见的共识情况进行综合。本文提出基于二分图网络投影算法的专家意见综合方法,并对传统的二分图边权重单一的不足进行了改进,使其减少投影过程中的信息损失,以满足复杂条件的分析需要。实证表明,改进之后的算法能够准确地反映不同专家间的意见相似度以及单个专家在群体中的独立性。最后,利用非线性映射技术,以可视化方式直观显示专家意见的共识情况。

To solve complex system problems needs not only data, information and model, but more importantly, opinions from experts. However, experts＇ opinions are quite different from one another, so it is necessary to synthesize the opinions to get a consensus. Thus, a bipartite network is presented in this paper to synthesize experts＇ opinions. Given the traditional bipartite network＇ s disadvantage that the edge weights are single and thus couldn＇ t meet requirement for complex problems, an improved projection algorithm for reducing in- formation loss is proposed. Demonstrative research indicates that new algorithm can precisely depict the relationship between different ex-perts and the independence of an individual expert in group. Finally, a nonlinear mapping technique is used to visually display the result of expert consensus.