关键词: 指标权重；群体DEMATEL；三维密度算子；评价标度；群体共识度

Abstract: In the face of some complex multiple-index comprehensive evaluation problems, the casual influence relationships among indexes are universal phenomenon and make some distinct effects for the rank of alternatives, and therefore decision-making trial and evaluation laboratory (DEMATEL) is given to eliminate these adverse effects at the process of evaluating the alternatives. DEMATEL is a widely applied method to determine the index weights by analyzing the casual influence relationships among indexes, and has attracted considerable attention from domestic and overseas experts and scholars in recent years. However, So far the methods of index weights based DEMATEL are mostly applicable to individual decision making, and do not consider evaluation scales are inconsistent under group decision making. To overcome this kind of defects, a new index weights calculation method of group DEMATEL based on three-dimensional density operator is proposed. Firstly, a transformation function of different evaluation scales is defined, which can convert inconsistent evaluation scales to single evaluation scale under group DEMATEL. Secondly, a cluster method of group DEMATEL matrix is given, on this basis, it is aggregated by three-dimensional density operator. Finally, the centre degree and cause degree of index is determined by using DEMATEL method, and the index weights is calculated. One exact example is used to illustrated the feasibility and validity of the proposed method. The results show that the index weights is reasonable and reliable as it can solve group decision making’s inconsistent evaluation scales and consider group decision making’s consensus degree. The study of the traditional DEMATEL method is expanded under individual decision making, and an important reference and guidance is provided for solving the problem of DEMATEL under group decision making in this paper.

Key words: index weights; group DEMATEL; three-dimensional density operator; evaluation scale; group consensus degree