Research On Group DEMATEL Decision-making Method Based On A New Scale In The Context Of Expert Interaction

As a kind of complex system factor analysis algorithm,decision making trial and evaluation laboratory(DEMATEL)is widely used in practice.Based a given evaluation scale,experts are invited to evaluate direct influence relationships between factors by mobilizing their expertise and prior experience.Then the direct influence matrix will be got,and the key factors will be determined through matrix operations.The rationality of the selected evaluation scale is one of the important premises and foundations to make DEMATEL decision-making result reasonable.However,by viewing the researches,the existing scale without exception are unable to describe the negative direct influence relationships between factors,and the negative influences are expressed and processed as positive ones.It is obvious that the actual influence relationships between factors can't be reasonably described.In addition,limited by experts' expertise and personal ability,the judgment results given by experts are often incomplete as the decision-making environment becomes increasingly complex,and the reliability of experts' judgment information also varies to some extent,which further leads to the failure of expert judgment consensus to achieve the desired effect.However,from the existing studies,on the one hand,most scholars have not considered the incompleteness of experts' judgment information,or have not provided the inference mechanism of incomplete information.It is obvious that the missing information is not necessarily worthless,and the neglect method is biased.On the other hand,in group DEMATEL decision-making,most scholars lack the consideration and evaluation of the reliability of experts' judgment information,and assume that experts reach a consensus on the judgment result and directly integrate the experts' judgments,which results in poor reliability of group judgment result and low expert consensus,thus affecting the quality of decision-making.In order to overcome these defects,this paper extends traditional DEMATEL evaluation scale,and defines the new scale which considers both positive and negative influences between factors.Drawing on the relevant ideas of knowledgeintegration theory,an experts' information integration method is proposed based on aggregation hierarchical clustering theory to improve the reliability of experts' judgment and group consensus degree.Combining the theory of trust relationship in social network,the mechanism of inferring incomplete information is proposed which is based on trust relationship between experts and the similarity degree of judgment.Finally,the improved DEMATEL method proposed in this paper is applied to a water supply company to analyze the influencing factors of non-revenue water.The case application shows that the method proposed in this paper is effective and feasible.