دانلود رایگان مقاله انگلیسی چارچوب عادلانه برای تصمیم گیری چند معیاره - الزویر 2018

ABSTRACT

As the determination of criteria weights is important for multiple criteria decision making, a number of attempts have been made to assign weights to criteria. However, whether criterion weight assignment is fair to each criterion and to each alternative is rarely taken into account. To address this issue, in this paper, we propose a fair framework in the context of the evidential reasoning approach, which is a type of multiple criteria utility function method. In the fair framework, two strategies are prepared for a decision maker to choose, which are the superior strategy and the inferior strategy. To achieve the objective in line with the selected strategy, two levels of fairness including the fairness among criteria and the fairness among alternatives are defined based on the performances of alternatives on each criterion. By following the two levels of fairness defined, two optimization models are constructed successively to generate possible sets of fair criterion weights. With a view to making all possible sets of fair criterion weights treated in generating a solution, they are incorporated into another optimization model constructed to generate the minimum and maximum expected utilities of each alternative, by which the solution is made with a decision rule preferred by the decision maker. A supplier evaluation problem is analyzed to demonstrate the applicability and validity of the fair framework.

6. Conclusions

The weight of a criterion is an important concept in MCDM that reflects the impact of the individual assessment of the criterion on the overall assessment. Appropriate or sound criterion weight assignment is crucial to making rational trade-offs among all criteria in MCDM methods. To focus on the assignment of weights to criteria, a large amount of research has been conducted. Although so, whether the criterion weight assignment is fair to each criterion and to each alternative is rarely taken into account in existing studies. To address this problem, a fair framework in the context of the ER approach was explored in this paper, which mainly includes two levels of fairness, i.e., the fairness among criteria and the fairness among alternatives. In the fair framework, two strategies were firstly provided for the decision maker to choose with the consideration of the decision problem, which are the superior strategy and the inferior strategy. After the choice, the fairness among criteria was defined and used to construct an optimization model to help each alternative achieve the objective in line with the selected strategy to the maximum extent. Based on the results derived from solving the optimization model, the fairness among alternatives was defined and used to construct the other optimization model to produce fair criterion weights in line with the selected strategy. By following the idea of treating all possible sets of fair criterion weights in line with the selected strategy, another optimization model was constructed to generate the minimum and maximum expected utilities of each alternative, which were used to generate a solution to the decision problem considered by exploring the decision maker’s behaviors or what decision rule is preferred by the decision maker.