Research On Methods Of Group Decision Making With Incomplete And Uncertain Information

Group decision making (GDM) is to aggregate different decision makers’ preferences into a consensus or comprise preference of the group according to some rules, which exists widely in socio-economic activities, politics activities and daily life of human beings. In recent years, the research on GDM has received increasing attention from scholars all over the world. However, there are some weaknesses in the existing literature. For some complex GDM problems, due to the uncertainty of decision environment and the difference of culture and education background among decision makers, decision makers tend to provide uncertain, incomplete or heterogeneous evaluation (preference) information. With the rapid development of information technology, more and more decision makers can participate in the GDM process. As a result, some major group decision making problems may involve different decision organizations. Therefore, one interesting question is how to represent and aggregate the evaluation (preference) information of decision organizations. To deal with the aforementioned problems, this thesis focuses on methods of GDM with incomplete and uncertain information. The main research work of this thesis is summarized as follows.(1) Multi-attribute GDM problems based on multi-granularity linguistic information are investigated. First, an approach is developed to deal with multi-attribute GDM problems with incomplete weight information under multi-granularity uncertain linguistic environment. For this type of decision making problems, some optimization models, which aim to minimize the deviation between the opinion of each individual and that of the group, are established to derive the collective evaluation of alternatives. Second, this thesis studies the unification problems of multi-granularity distribution linguistic assessments and develops two distribution linguistic power aggregation operators which can take the relationships among the input arguments into account, based on which two approaches to multi-attribute GDM based on multi-granularity distribution linguistic assessments are proposed.(2) Decision making methods based on uncertain linguistic preference relations are inves-tigated. First, the consistency of uncertain2-tuple linguistic preference relations is defined, and then two algorithms are developed to estimate the missing elements for an incomplete uncer-tain2-tuple linguistic preference relation, which can be used to deal with GDM problems with incomplete uncertain2-tuple linguistic preference relations. Afterwards, this thesis studies the individual consistency and group consensus problems for uncertain2-tuple linguistic preference relations. To this end, the individual consistency index and the group consensus index are de-fined, based on which two algorithms are developed to improve the individual consistency and group consensus for GDM problems based on uncertain2-tuple linguistic preference relations.(3) This thesis also investigates GDM problems based on heterogeneous incomplete pref-erence relations. For GDM problems with heterogeneous incomplete uncertain preference re-lations, this thesis defines the group consensus index and the collective consistency index, and then establishes a bi-objective optimization model which aims to minimize the two indices to derive the priority weights of alternatives. Moreover, this thesis makes a study on GDM with in-complete hesitant preference relations, including hesitant fuzzy preference relations and hesitant multiplicative preference relations. For this type of GDM problems, some formulae which can derive an individual priority weight vector from an incomplete hesitant preference relation are proposed based on the logarithmic least squares method. Based on the formulae, an approach is developed to deal with GDM problems in which different decision organizations are involved.(4) An approach is proposed to deal with GDM problems with ordinal intervals and un-certain weight information of decision makers. In this approach, a multi-objective optimization model which aims to maximize the deviation of each decision maker’s judgments and the consis-tency among different decision makers’ judgments is established to obtain the weights of deci-sion makers. After that, the thesis extends the VIKOR approach to GDM problems with ordinal intervals and proposes a decision analysis method.To summarize, the research work presented in this thesis enriches the theory and methods of GDM under complex uncertain environment. The proposed models and algorithms can be applied to deal with practical decision making problems, such as enterprise policy development, supplier selection and new product planning.