不完全情報を考慮した集団ANPの提案

抄録

In this paper, we propose an ANP framework for group decision-making problem with incomplete information. The proposed method consists of three parts, namely, aggregation of weights of alternatives, aggregation of weights of criteria and evaluation of global weights. In aggregating weights of alternatives, we calculate group weights from individual weights, which have been calculated by some AHP, by using a quadratic programming problem. In aggregating the weights of criteria, we calculate the weights for evaluating an alternative from the matrix, which is geometric average of pairwise comparison matrices of decision makers only who evaluate the alternative. Then the weights of criteria may be different in each alternative. Hence, we use the supermatrix method of ANP for evaluating global weights. A small numerical example shows that our proposed framework enables us to obtain the global weights of a group decision-making problem with incomplete information.