Development multiple criteria decision support system for resolving conflicts among group members

Abstract

Decisions with multiple objectives are quite prevalent in government, military, industry and other organizations. Researchers from a wide variety of disciplines such as mathematics, management science, economics, engineering and others, have contributed to the solution method for Multiple Objective Decision Making (MODM) problems. In practice, the problems formulated as MODM models have a feature that multiple participants take part in the problem solving process. It would be reasonable to assume that the public sector analyst simultaneously serves several participants with their conflicting viewpoints. However, unlike MODM solution methods, very few interactive approaches have been studied in group decision with multiple objectives to find a best compromise solution. Further, the so called complete conflict case which two participants show symmetric preference structures with respect to one criterion, has not been investigated. Moreover, Multiple Criteria Decision Support System (MCDSS) to support group decision with multiple objectives fell behind. The main concern of this research is to develop MCDSS which supports group members to derive a properly satisfied final solution to multiple objective linear programming problems. The proposed method consists of the interactive algorithms to search a best compromise solution in group decision environments, which would be divided into general conflict case and complete conflict one, and of Multiple Objective Multiple Decision-Maker Problem Solver (MOMDPS) to support the progress of achieving the best compromise solution. The basic algorithm for the general conflict case adopts the local approximation of participants preference structures by employing various local proxy preference functions and interaction with them. Through local approximation of utility function, the bias and information loss resulting from global approximation adopted in earlier researches can be reduced. The algorithm for the complete confli...