社会网络多属性群决策中的策略权操纵模型及应用研究

With the development of information and network technology, a new type of social network multiple attribute group decision making (SNMAGDM) problem has to be faced in the process of evaluation and decision making. In reality, a decision maker can strategically set attribute weights to obtain his/her own interests, which is called the strategic weight manipulation of the SNMAGDM in this project. First, on the basis of refining the characteristics of attribute weights in social network, we construct an optimization-based strategic weight manipulation model with minimum social network adjustments in SNMAGDM, to obtain the manipulator’s desired ranking of alternatives, and design an algorithm to solve the proposed model by analyzing its complexity of computation. Then, a ranking range model based social network is proposed to manage the strategic manipulation behavior, and a comparative analysis on the performance of defending against strategic weight manipulation for different aggregation functions is provided. Meanwhile, due to the different decision makers may use heterogeneous preference representation structures to express their preferences, the strategic weight manipulation and ranking range models in heterogeneous SNMAGDM are developed, which can further improve the existing theories and methods of SNMAGDM. Finally, some detailed simulation experiments are designed to justify our proposal, and the theoretical results can provide a decision support to identify and manage the strategic manipulation behaviors in realistic SNMAGDM.

随着网络信息技术的快速发展，社会网络多属性群决策成为评价与决策中的一个新兴研究问题。项目通过设计策略的属性权，使得决策结果符合自身利益为出发点，对社会网络多属性群决策中的策略操纵问题展开研究。首先，基于社会网络和属性权重的交互影响特征，建立最小调整社会网络的策略权操纵模型，以此得到操纵者期望的方案排名，通过对模型的计算复杂性进行分析，设计策略权操纵的求解算法。进一步，从防操纵视角，建立基于社会网络的防操纵机制模型，比较不同集结函数防范策略操纵的性能。同时，结合现实评价与决策问题中，不同决策者偏好表达的差异性，研究异质环境下基于社会网络的策略权操纵模型及防操纵机制，丰富社会网络多属性群决策的理论与方法。最后，通过实验仿真对理论模型进行检验与修正，并将其应用到现实社会网络多属性群决策中的策略操纵行为的识别与管理当中。

随着信息技术和网络的快速发展，社会网络多属性群决策日益受到重视。现实的社会网络多属性群决策过程中，经常有决策者是不诚实的，他/她试图通过设定策略的属性权使得决策结果符合自身偏好，称为社会网络多属性群决策中的策略权操纵。本项目以策略操纵为基本点，围绕以下问题展开研究：（1）从策略操纵视角，建立社会网络多属性群决策的策略权操纵模型；(2)从防操纵视角，建立方案排序范围模型；(3)理论结果的应用。首先，建立最小调整社会网络的策略权操纵模型来设定策略的属性权，得到操纵者期望的方案排名。然后，由于方案排序范围可以用来测量属性权重动态变化下方案排名的下界和上界，以及属性权重与社会网络的交互影响特征，建立了基于社会网络的方案排序范围模型。最后，讨论了理论模型在项目评级和大学排名中的应用，并设计了详细的仿真实验，其结果验证了所提出方法的有效性。