Research On The Formation Of Social Network Cohesion Based On The Core Periphery Structure

With the continuous development of the Internet and the continuous emergence of a variety of social platforms,the relationship between people is getting more closer,and the ways of communication and interaction are becoming more and more diversified.In order to better analyze and understand these social phenomena,social network analysis has gradually become a research hotspot in the computer field,that is,the use of computer technology to study the relationship between social groups and a series of social phenomena caused by the interaction between group members.The research on the dynamic behavior of social groups,such as the emergence of social groups and the formation of social customs,has always been an important topic in computer science and artificial intelligence.It is necessary in many cases to unite all the members of a group to form a cohesive group.This is of great significance for social cooperation,social customs and other related research.Therefore,it is necessary to study the relationship between social cohesion and the dynamic behavior of social groups,and this overall point of view naturally needs to introduce the idea of game theory.The formation of coalition is an important research direction in cooperative game,which is usually called the coalition game.The model of coalition game is often used to solve the problems of resource allocation,community formation and so on.However,most studies fail to consider the formation of a stable large alliance(that is,group members joining a same coalition)by adjusting the network structure to promote the formation of the cohesion of the entire social group.This paper studies a dynamic behavior model of social groups based on cooperative games—The Influence Game model.Based on the traditional network structure,social cohesion is defined based on this model.This paper points out that when all the nodes in a network choose to join the same coalition to form a stable coalition,the network has social cohesion.In addition,through the study of the network structure with social cohesion,this paper proposes a recommendation mechanism to promote the formation of network social cohesion(ie,the core periphery algorithm).This algorithm provides a reference for the formation of network social cohesion,and analyzes the proposed algorithm through simulation experiments in artificial generated network and real networ.The main research contents of this paper are as follows:(1)This paper proposes a reasonable and easy-to-understand cooperative game model—the influence game model.This model not only takes into account the factors that affect the income of individual members in the coalition,but also considers the expenditures required by individuals to join the coalition.In this model,it is believed that the individual's ultimate income is related to their expenditure and income,which is also the basis for individuals to choose to join the coalition during the formation of the coalition.(2)Based on the influence game model,this paper discusses what kind of network structure has social cohesion from the study of traditional network structure.In this paper,it is proved theoretically that complete graph,normal core periphery network(Normal Core Periphery,referred to as NCP)and approximate normal core periphery network(Approximate Normal Core Periphery,referred to as ACP)have social cohesion.This provides a theoretical support for the later research on how to form a cohesive network structure.(3)Based on the core periphery structure(Core Periphery,referred to as CP),this paper proposes an algorithm for how to form a network structure with social cohesion.That is,the network is divided into the core layer and the periphery layer,and then the links between the nodes are increased by increasing the edges between the core nodes and the periphery nodes,so as to promote the formation of social cohesion of the network.Based on the consideration of individual characteristics and groups,this paper proposes a core periphery algorithm based on important nodes(referred to as CPIN),a core periphery algorithm based on maximum clique(referred to as CPMC),and a core periphery algorithm based on spectral clustering(referred to as CPSC).(4)In order to verify the effectiveness of the proposed algorithm,this paper compares the proposed algorithm with the previous algorithm in 3 different synthetic networks(scale-free network(BA),random network(ER),rich club network(Rich-Club))and 11 real networks.The performance of the algorithm proposed in this paper is analyzed and summarized through simulation experiments.