A new method research for knowledge-match and trust-based large-scale group decision making with incomplete information context

Abstract

In the process of large-scale group decision making (LSGDM), probabilistic linguistic term set (PLTS) is an useful tool to represent the preferences of expert. There is a common case that experts tend to provide incomplete preferences due to various reasons. However, previous methods which cope with the missing values never took experts′ level of cognition over alternatives and attributes into account. In reality, because of limited knowledge reservation and the complexity of decision problem, experts have diverse familiarity with each scheme and attribute. For handling the defect, we propose a novel method to fill missing preference values, based on the combination of knowledge-match degree and trust degree of experts providing reference information. We obtain the knowledge-match degree through the accuracy and reliability of preference as well as the trust degree through social network analysis technology (SNA), and use the probabilistic linguistic weighted average operator (PLWA) to integrate the referential values into preferences of the missing expert. Moreover, to solve the consensus problem at minimal cost, a consensus model based minimum adjust is developed in which the consensus degree of identified elements are all lowest at three aspects including decision matrix, internal experts and intra-group. On the basis of the trust relationship, revising the preference with low consensus guarantees regulated experts′ real aspiration. In addition, a new approach to measure the weight of sub-group is proposed in the light of trust in-degree which considers the reliability of experts in the same subgroup.The feasibility and validity of the LSGDM method are tested by using a numerical example and comparing with other methods.