1607. Consensus reaching process in large-scale group decision making based on bounded confidence and social network
Invited abstract in session MD-4: EJOR: policy, facts and highlights in stream OR Journals, stream OR Journals.
Monday, 14:30-16:00Room: Rupert Beckett LT
Authors (first author is the speaker)
| 1. | Yanhong Li
|
| Chengdu University |
Abstract
Group decision making (GDM) consists of a group of experts to make the satisfactory decision, it plays a vital impact on people’s activities. In GDM, Different experts often express different opinions, and if they fail to reach a consensus, it often leads to decision failure and affects the results of the decision. Consensus reaching process (CRP) is a dynamic and interactive method used to reach a group decision. Now that social networks and mobile internet are prominent features in daily life, more experts are able to take participate in decision making in the network and their opinions are influenced by others in the decision-making process. Therefore, how to use the difference of opinions and the relationships between experts to promote the consensus is an important issue. When the number of experts participating in the decision-making process is over 20, it involves large-scale group decision-making (LSGDM).
In these cases, there are the following issues: (1) Existing studies may ignore experts’ unique knowledge background, interests and social relations. As a key factor in information aggregation, it is more reasonable to use experts’ social network relations to determine the weight. (2) When experts do not reach a consensus, previous studies usually assume that the suggestions generated by the feedback mechanism will always be accepted by experts. However, providing feedback does not mean that experts will fully accept different suggestions. If the suggestions generated by the feedback mechanism are far from the experts’ opinions, the experts will tend to reject the suggestions rather than adjust the opinions. In addition, the known group consensus model ignores the trust relationship between experts, so it can not directly guide the group decision-making behavior under the social network. (3) How to make full use of the trust relationship and opinion similarity of experts to improve the decision-making efficiency in LSGDM.
Motivated by the challenge to overcome the above limitations, we assume that the experts tent to accept advice from others they trust and the experts tend to accept advice if the distances between these opinions and their own opinions are less than or equal to a certain confidence value. Therefore, we proposed a CRP approach in LSGDM based on bounded confidence and social network. The approach studied group consensus from social network relationships and differences in opinions and builds the feedback mechanism to help experts reach a consensus. The main contributions and novelties were obtained as follows: (1) A fast unfolding algorithm based on the social relationships between experts is applied to divide the large-scale into the small sub-clusters, which can improve decision efficiency; (2) An algorithm for social network analysis is proposed to determine the weights of experts; (3) A new feedback mechanism based on bounded confidence and social network is designed to promote experts reach a consensus. Compared to the existing methods, the approach directly considers the social relationships between experts and bounded confidence. Thus, it avoids to consider the factors that affect the decision-making results only from social networks or bounded confidence. We also compare our results with different methods using qualitative and quantitative analyses. The simulation analyses are used to show the efficiency of the proposed consensus reaching model approach, the results show that when the social network relationship of experts is static, the degree of experts adhering to their own opinions and bounded confidence in others’ opinions will affect the decision-making results.
Keywords
- Group Decision Making and Negotiation
- Decision Analysis
Status: accepted
Back to the list of papers