Research On The DEMATEL Decision-making Method Of Complex System Group Based On The Difference Expert Evaluation Scale

In a knowledge economy era with rapidly changed science and technology,organizations are facing many problems caused by the complexity and uncertainty of environment and the fierce competitive.There arose an important research subject as how to analyze the key success factors taking actions to improve the performance of organizations.Introduced by the Geneva Research Center of Batelle Memorial Institute in the 1970 s,Decision-making trial and evaluation laboratory(DEMATEL)is a widely applied method used in the research of complicated socioeconomic system issues.The method can mirror the causal interactions of factors and identify the critical factors in tabular and graphical form.Due to its universality and simplicity,DEMATEL has been applied in many situations.Because of the complexity and uncertainty of the socioeconomic system,a group of experts are always needed in the information gathering phase of the DEMATEL method.In order to enhance the quality of information,three aspects must be considered in this phase:(1)The evaluation scales when experts possess different granules of knowledge.(2)The completeness of information with complex and uncertain system.(3)The expression of judgments of the experts with hesitant.As such,this dissertation focuses on the amended method of group DEMATEL with multi-granular evaluation scales,and provides detailed algorithm according to the following situations:Firstly,the group experts provide completed linguistic preference judgments with multigranular evaluation scales.The proposed method applied 2-tuple linguistic model and extended hierarchical linguistic model to unify the multi-granular information.Linguistic evaluations from different experts are transferred into 2-tuple linguistic initially.An extended hierarchical linguistic is construct with linguistic scales provided by experts and an added level contains all the modeling points of previous levels.Then the unification between different levels can be based on transferred operators between levels,and the unified information can be aggregated with average operator of 2-tuple linguistic for the group DEMATEL processing.Secondly,the group experts provide completed preference judgments with hesitant fuzzy linguistic term sets and multi-granular evaluation scales.The dissertation proposes an aggregation algorithm of multi-granular information based on a comparison of the possibility degrees of HFLTS.It also amend the DEMATEL method with a novel perspective on the influence between factors and the aggregation algorithm of multi-granular information.Two possibility thresholds are defined to indicate the user's preference and generate the group consensus information for further analysis.Thirdly,the group experts provide incomplete preference judgments with multi-granular evaluation scales.In order to aggregate the incomplete and multi-granular information,the influences between factors are considered as an alternative to be compared in a total ordered sequences with different lengths.Group ranking approach for ordinal preferences based on group maximum consensus sequences is applied to aggregate the sequences of influences.Then the consensus scores generated by a group consensus-ranking model are regard as elements of group initial direct influence matrix.So the group DEMATEL method can be processed based on the consensus scores direct influence matrix.Fourthly,the group experts provide incomplete preference judgments with hesitant fuzzy linguistic term sets and multi-granular evaluation scales.There is a hypotheses in the DEMATEL method that the influence of factor on itself is considered as none for all factors.Based on the hypotheses,distance measurement on hesitant fuzzy linguistic term sets can be applied on the unification of multi-granular evaluation information.The loss information are classified for completion processed.Then a distance measurement matrix of direct influence is defined to aggregate the information.Finally,an interactive aggregation algorithm is applied to generating group consensus direct influence matrix.In a conclusion,the dissertation amends traditional group DEMATEL considering the multi-granular evaluation scales,especially provides detailed algorithm from the viewpoint of incomplete information and hesitant fuzzy linguistic term sets.It also applies the proposed methods into actual case and four illustrate examples in each chapter.The feasibility and effectiveness are proved during the process of examples.