Abstract:Nowadays, more and more decision-makers (DMs) are engaging in group decision-making (GDM) within certain social relationship networks. Therefore, understanding how to leverage differences in DMs' opinions and social relationships to promote consensus in large-scale group decision-making (LSGDM) is an important issue. This study proposes a bi-level consensus model for LSGDM in social networks, well considering social influence to achieve the objective of minimum cost of the upper-level mediator and maximum satisfaction of lower-level subgroups. Firstly, the Louvain algorithm is employed to reduce the dimensions of LSGDM, segmenting DMs in social networks into distinct subgroups in a directed graph. Then, a dynamic opinion experiment based on the Friedkin-Johnsen model is utilized to assess the confidence levels of subgroup members and enhance opinion coherence within subgroups. Operating at the subgroup perspective and adopting a dual-layer framework, this study establishes the minimum cost maximum satisfaction consensus model (MCMSCM) to better balance the objectives between the upper and lower levels. Furthermore, a bi-level nested algorithm, based on genetic algorithm, is employed to determine corresponding unit costs and adjusted opinions, thereby achieving consensus rapidly and effectively. The proposed methodology provides a robust tool for LSGDM in social networks. Finally, through an illustrative example accompanied by corresponding analysis, the rationality and effectiveness of this pattern are demonstrated.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to effectively utilize the opinion differences and social relationships among decision - makers to promote consensus - reaching when making large - scale group decisions in social networks. Specifically, the paper proposes a two - layer consensus model, aiming to minimize the cost of upper - level coordinators while maximizing the satisfaction of lower - level subgroups. This model specifically takes into account social influence factors, reduces the decision - making scale by using the Louvain algorithm, divides decision - makers into different subgroups, and uses dynamic opinion experiments to evaluate the confidence levels of subgroup members and enhance the opinion consistency within subgroups. Finally, the paper proposes a two - layer nested algorithm based on the genetic algorithm to determine the corresponding unit costs and adjusted opinions, thereby achieving consensus quickly and effectively. ### Main contributions of the paper 1. **Two - layer consensus optimization model**: A two - layer consensus optimization model (MCMSCM) based on minimum cost and maximum satisfaction is proposed. This model adopts a two - layer perspective method and effectively balances the goals of upper - level coordinators and lower - level subgroups. 2. **Application of community detection algorithm**: In a directed social network, the efficient Louvain algorithm is applied for community detection to divide the original decision - makers into different subgroups for dimension reduction. In addition, the modularity and modularity gain of the Louvain algorithm in a directed graph are defined. 3. **Consideration of social influence**: In the context of social networks, not only the satisfaction of decision - makers and the limits of opinion adjustment are considered, but also social influence. Based on dynamic opinion learning, the influence of others' opinions on decision - makers' self - opinions is introduced, and the degree of influence of this influence on decision - makers' self - confidence is measured. ### Model construction #### 3.1 Dimension reduction and Louvain algorithm - **Louvain algorithm**: Identify closely - connected community structures in complex networks by optimizing modularity. The algorithm is divided into two stages: in the first stage, each node is initially regarded as an independent community, calculate the modularity gain of moving the node to the community to which its neighbor belongs, and select the community that maximizes the modularity gain; in the second stage, merge the communities obtained in the first stage into new nodes to form a new directed graph, and repeat the calculation of modularity gain until the modularity no longer increases. #### 3.1 Weight allocation strategy and self - confidence - **Mixed weight allocation strategy**: A mixed weight allocation strategy is adopted within and between subgroups, and the final weight is determined by the social influence weight and the initial average weight. The social influence weight reflects the trust and influence of decision - makers in the social network, while the initial average weight represents the uniform distribution of weights among individuals or subgroups. - **Dynamic opinion experiment**: Conduct a dynamic opinion experiment in the initial social network. After subgroup members express their initial opinions, these opinions are transparently displayed, and after reflection, they express their final opinions. The final opinion is determined by the initial opinion and social influence together and is affected by the self - confidence of members. #### 3.3 Construction of two - layer model - **Minimum cost model of upper - level coordinators**: Based on the consensus model at the subgroup level, calculate the self - confidence weight of each decision - maker. There are few or even no social connections between subgroups. - **Maximum satisfaction model of lower - level subgroups**: Determine the unit cost and adjusted opinions through a two - layer nested algorithm to achieve rapid and effective consensus - reaching. ### Conclusion The two - layer consensus mechanism proposed in the paper provides a powerful tool for large - scale group decision - making in social networks. By comprehensively considering social influence and multi - layer goals, it effectively promotes the achievement of consensus.

Facilitating large-scale group decision-making in social networks: A bi-level consensus model with social influence

A two-stage consensus model for large-scale group decision-making considering dynamic social networks

Social Network Group Decision-Making Method Based on Stochastic Multi-Criteria Acceptability Analysis for Probabilistic Linguistic Term Sets

A consensus method in social network large-scale group decision making with interval information

A Clustering- and Maximum Consensus-based Model for Social Network Large-scale Group Decision Making with Linguistic Distribution

Opinion Dynamics and Minimum Adjustment-Driven Consensus Model for Multi-Criteria Large-Scale Group Decision Making Under a Novel Social Trust Propagation Mechanism

The bi-level consensus model with dual social networks for group decision making

A Consensus Model for Large-Scale Group Decision-Making Based on the Trust Relationship Considering Leadership Behaviors and Non-cooperative Behaviors

A maximum satisfaction consensus-based large-scale group decision-making in social network considering limited compromise behavior

Consensus Reaching Mechanism with Parallel Dynamic Feedback Strategy for Large-Scale Group Decision Making under Social Network Analysis

Consensus reaching process with noncooperative behaviors in large-scale group social network environment

A Sentiment Analysis-Based Two-Stage Consensus Model of Large-Scale Group with Core-Periphery Structure.

An Influence Network-Based Consensus Model for Large-Scale Group Decision Making with Linguistic Information

Adaptive Consensus-Based Model for Heterogeneous Large-Scale Group Decision-Making: Detecting and Managing Noncooperative Behaviors

A large-scale group decision-making model with no consensus threshold based on social network analysis

How to process local and global consensus? A large-scale group decision making model based on social network analysis with probabilistic linguistic information

Large Group Decision-Making Method Based on Social Network Analysis: Integrating Evaluation Information and Trust Relationships

Consensus Reaching Process With Multiobjective Optimization for Large-Scale Group Decision Making With Cooperative Game

Managing noncooperative behaviors in large-scale group decision-making: Integration of independent and supervised consensus-reaching models

A minimum adjustment cost feedback mechanism based consensus model for group decision making under social network with distributed linguistic trust

A k-core decomposition-based opinion leaders identifying method and clustering-based consensus model for large-scale group decision making