Research On Social Network Cohesion Based On Hierarchical Structure

With the development of Internet technology and the rise of social networking platforms,people's ways of obtaining information have become diversified.People's social relationships have realized instant communication and resource sharing through the social networking platforms,which has greatly accelerated people's access to information and the spread of information process.Social network analysis has also spawned as a new branch in computer field.Social cohesion is an important concept in sociology,which is usually used to express the degree of unity in a social group.Social cohesion originally existed only in the study of sociology.But with the development of network structure analysis,more researchers have begun to explore the connotation of social cohesion through social network analysis.Combined with complex networks,many researchers have analyzed social cohesion from social network structure,and a new concept which called structural cohesion has emerged,and used to describe social cohesion.Structural cohesion is mostly measured based on social network connectivity.Depending on the research direction,the measures of structural cohesion are often different.Such as some research measures structural cohesion through the average shortest path of the network.These measures of structural cohesion have been validated successfully in related researches.However,these studies are based on the centrality of nodes,and ignore the class phenomenon in real society.Although the social network shows the feature of complex network,it often includes some more social concepts.Such as the propagation of social capital,knowledge and influence.Explore social networks only with complex network structures cannot visually observe these characteristics.Here we integrate the complex network structures and influence propagation efficiency of social networks to conduct an in-depth analysis of social networks.On the structural aspect,we use K-Shell decomposition to analyze the hierarchical structure of social networks,and explore the connectivity of social networks.From the perspective of influence propagation,we have used the IC model,which is a classic information propagation model.We use these to simulate influence propagation in social networks,and explore the relations between the connectivity of social network and the propagational efficiency of influence by observing the process of influence propagation in social networks.And the concept of average propagation intensity is proposed to describe the cohesion of social networks.In addition,we were inspired during the research process and began to consider the change mechanism of social network cohesion.Research shows that networks with densely connected relationships tend to be more cohesive.In the experimental part,we designed related algorithms based on the network hierarchy to establish new connection relationships in social networks.By comparing with traditional algorithms,the experimental results well show the superiority of our algorithm.This thesis focuses on the relations of network structure characteristics on the overall network propagation performance.The main research contents of this thesis are shown below:(1)From the perspective of influence propagation,this thesis focuses on the individual's location attributes and explores whether the centrality of the individual determines whether the individual plays an important intermediary role in the process of retransmission.(2)From the perspective of structure,this thesis explores the overall structural characteristics of the social networks,such as connectivity and core-edge structure.(3)Based on the hierarchical structure of the social network,the shell cohesion of the network is described in terms of the strength of the connection between the layers.The influence propagation process is simulated in the IC model.And the average strength of the network is used to verify the shell cohesion.On this basis,we further explore how to intervene the cohesion of social networks,and design SBDA algorithm to alter the cohesion.(4)In order to verify the rationality of the shell cohesion,a large number of experiments are performed in this paper.This article uses 10 generative networks,including small-world model,BA model,ER random model,fitness model and local-world evolving model;9 real social networks,including high school friendship network,email network,physicist networks,Facebook networks,and so on.The experiments show many important structural insights and validate the rationality of our proposed connection strategy.