Abstract: In this paper, a group is composed of dependent members who are supposed to be evaluated via multiple attributes and participation in determining these attributes' weights. In order to seek the attributes' weights in multiple attribute group decision making, an iterative algorithm is proposed, which starts with equal weights and solves a programming model with weight parameters for each group member to obtain a set of optimal weights. The optimal weights to each attribute decided by every group member are then averaged and used as the new weight parameters in modeling solving during the next iteration. This iterative procedure repeats until for each member, the consecutive weighted average values of all attributes converge within a specified small enough positive value. The optimal attributes' weights of the group are therefore obtained as the current parameter weights when the iteration terminates. Such a set of attributes' weights are jointly determined by all group members, in which sense the related results and decisions are supposed to be accepted and satisfied by all (or at least most) members. A numerical example and an application of R&D project selection are studied to demonstrate the feasibility and effectiveness of the proposed method.

Key words: group decision making, attributes' weights, interactive iteration