Research On Methods And Applications For Interactive Multiple Attribue Decision Making Under Uncertainty

Under the condition of information asymmetry, focused on arranging weights and information interaction of multi-attribute decision-making field, interactive multi-attribute decision problems are studied and modeled in this dissertation. This dissertation follows the thought of gradually advancing from the basic problem to synthesis modeling and to interactive synthesis modeling of multi-attribute decision-making. The information interaction of decision maker and alternative holders, and reutilization of decision rules in the process of interaction are depicted with model in the dissertation. And then a series of models are developed, such as sensitivity and robustness analysis, arranging weights of attributes, ranking and choice, extracting decision rules, and using decision rules. In detail problems in the following four aspects are addressed:(1) Some basic problems in multi-attribute decision filed are studied. Aimed at the need of sensitivity analysis, a new normalization method is developed based on the domain of attributes. A general algorithm is developed for the arranging weights based the idea of maximizing deviations. Based on the unique superior which dominance-based rough set theory is applied to multi-attribute decision filed, the determining weights method based on dominance-based rough set theory is developed.(2) A kind of linear programming model is used to carry out sensitivity and robustness analysis. Based on SAW operator, aiming at certain decision, with linear programming model, a new method is developed for sensitivity and robustness analysis. Different from the limitation that classic analysis method just changes two weights at the same time, the technological merit of new method is that we can take simultaneously changing all the weights of attributes into account. And so the method expands the research scope, and it helps to guide decision making with the analysis results.(3) The new idea is presented that we should study interactive multi-attribute decision under the condition of information asymmetry and provides corresponding solution to build an interactive multi-attribute decision model, which is based on compromise solution. Meanwhile, the problem of reusing decision rules is studied. In the realistic environment, information obtained by decision maker and alternative holders is asymmetrical. However in the existing literature, all the interactive decision models were built on the view of symmetric information. The condition of information asymmetry is selected for studying the interactive multi-attribute decision problem, which expands the research scope and helps to push the application of the study conclusions for practical problem forward. On the other hand, based on multi-graded dominance criterion, a new multi-attribute decision model P2DRS is developed. Its core is that the model can change the classic decision table into pairwise comparison table, which can avoid information loss caused by conventional much too simple changing method.The model can easily deal with two problems, and one is that alternatives with the same net flow value can not be completely ranked, the other is that the pre-existing decision rules can not resolve certain new cases.(4) The models proposed in this dissertation are used to deal with practical cases such as public bidding, purchasing, and multi-attribute auction. And then these cases and their conclusions prove that the methods proposed in the dissertation are useful and valid.