The Research On Comprehensive Evaluation Methods With Incomplete Information

The comprehensive evaluation problems could be found abroad in the areas of sociology, economics and management. The research on the theory and method of comprehensive evaluation has an extensive prospect. Duing to the complexity of the objective world, the limitation of people’s ideation and knowledge, people will often encount the comprehensive evaluation problems with incomplete information in practice. For example, the evaluation parameter presenting as linear inequation, being short of segmental evaluation information, the incomplete judgement matrix, the distributed structure of evaluation information and preference information with the subset of alternatives. They are all could be found in comprehensive evaluation problems with incomplete information. The conventional methods could not be used directly to the comprehensive evaluation problems with incomplete information. So it is necessary to do research on the methods of comprehensive evaluation problems with incomplete information.Dempster-Shafer theory is used to deal with the comprehensive evaluation problems with incomplete information frequently. So it is important to know that how does the Dempster-Shafer theory being used to deal with the problems, what kind of problems with incomplete information could be done with Dempster-Shafer theory, and if the existent methods based on Dempster-Shafer theory is proper. The determination of attribute weights is a nuclear problem of comprehensive evaluation, so it also should be considered under the condition of attribute weights with incomplete preference information. The comprehensive evaluation problem could be extended to group evaluation environment and time series environment, and it is necessary to consider the management of incomplete information under group evaluation environment and the time series environment. Corresponding to the affluent production of conventional comprehensive evaluation, there are many work need to be done with the comprehensive evaluation with incomplete information. The main research works of this dissertation are as follows:(1) The research of modeling of comprehensive evaluation with incomplete information. For multi-attribute comprehensive evaluation problems with segmental evaluation information, this dissertation analyzes the appropriateness of modeling in existing literature and proposes an improved model with Dempster-Shafer theory. The resolution method, decision-making rule and excellent characteristics of this improved model are then studied.(2) The research of determination of attribute weights with comprehensive evaluation with incomplete information.①a kind of multi-attribute bargaining evaluation problem is considered which combines the bargaining evaluation and self-profit evaluation. The problem is described as attribute weights with linear inequation, and a multi-variable Induced Ordered Weighted Averaging (MIOWA) operator is proposed to solve the problem.②The determination of attribute weights is studied under group evaluation with segmental attribute evaluation information. And a method is proposed based on Dempster-Shafer theory which could deal with the segmental attribute evaluation information directly. This is distinct with the means of incomplete judgement matrix and could tackle the situation of unacceptable incomplete judgement matrix. The information synthesizing process is deduced, so it could be treated in program and efficiently.(3) The group evaluation problem with incomplete information is considered in the form of incomplete judgement matrix, the distributed structure of evaluation information and preference information with the subsets of alternatives. The relative theory of OWA operator is used to solve the problems. Combining the "most of fuzzy linguistic quantifier, the incomplete preference informations of expert group are integrated with the weighting vector, and the integration process embodys the "majority rule" of group evaluation.①For problems in the form of incomplete judgement matrix with linguistic preference information, an improved group evaluation method is proposed. The conversion formulas are given for the pretreatment of experts’preference information, and two induced ordered weighted averaging aggregation operators for two-tuple linguistic preference information are defined for the integration of experts’preference information.②For group evaluation problems in the form of the distributed structure of evaluation information, an information integration method is presented while the experts’ weight coefficients are unknown. Based on the Evidential Reasoning approach by Yang J B et.al, the information integration process is deduced with the weighting vector, which is used to adjust the basic belief assignment of envidence from every expert.③For group evaluation problems in the form of preference information with the subsets of alternatives, an information integration method is presented while the experts’weight coefficients are unknown. The improvement work is done to DS/AHP, and the collective distance measure of the evidence from individual expert is defined, then the integration of experts’preference information is done combining the weighting vector and "most of fuzzy linguistic quantifier.(4) For a dynamic comprehensive evaluation problem with incomplete attribute preference information, a model is set up with the attribute weight in the form of linear inequation. During the determination of attribute weights, the "difference drive" and "function drive" are used. The time-orness index is used to reflect the time factor in dynamic comprehensive evaluation, and the meaning of time-orness index is analyzed further. The dynamic comprehensive evaluation model is used to check the public expenditure performances of fourteen cities in Liaoning province.Finally, the whole dissertation is summarized, as well as future research problems.