Research On Community Detection And Cross-community Information Dissemination Model Based On Multi-dimensional Feature Fusion

With the rapid development of multimedia social platforms,online social networks have become more complex and larger in scale.In these social platforms,the communication of friends gradually forms community groups,and these groups become the media of information dissemination.At the same time,the process of information dissemination It will also in turn affect the results of community testing.Therefore,it is becoming more and more important to conduct community detection and information dissemination on online social networks.Using a more reasonable method to study the phenomenon of information dissemination caused by community structure and ensuring the accuracy of the results as much as possible is of great significance for exploring the scope of information dissemination.In the current social network research work,most of the research on community detection and information dissemination starts from the topological network.By exploring the impact of topology structure on community division and information dissemination,the current model has achieved certain results,but there are still The following questions:1)Most algorithms currently only consider a single topology information for community division.However,in real life,each node has its own characteristics and attributes.Only relying on a single topology information will reduce the accuracy of community detection results.What this paper aims to solve is how to comprehensively balance the use of topology-attribute information to improve the quality of community detection.2)At present,most of the information dissemination process considers the global network,and the overlay network is propagated based on high-influence nodes,but information is hindered from spreading across community structures over time.However,the dissemination of information between community structures will be hindered over time.At the same time,only considering a single relationship will also affect the dissemination of information.Considering the influence of strong and weak relationships,the community structure is used to maximize the scope of information dissemination.Aiming at the above problems,this paper designs a universal algorithm based on related theories and calculation methods.Firstly,starting from node attributes,a multidimensional feature fusion community detection algorithm is proposed.Considering the different influences of node relationships on user communication in the community structure network,a cross-community information diffusion model based on strong and weak relationships is proposed.1)In the problem of multi-dimensional feature fusion community detection in social networks,first,filter the attribute information in the network,that is,filter out homogeneous attributes according to the attribute-structure influence degree and information entropy,and then use the attributes to enhance the network.Then in the middle stage,the topology information and attribute information are fused,and finally a new objective function is obtained,and community detection is performed on the fused network.2)In the cross-community information dissemination problem based on strong and weak relationship,first,model from the community structure and node relationship,use the community structure and node relationship to determine the bridge link,further obtain important nodes and driving nodes required for propagation according to bridge links,and then formulate new information propagation rules.Based on the above two aspects,an information dissemination model suitable for cross-community is obtained,and finally,the maximization of the dissemination range is explored on the basis of information dissemination.3)By comparing experiments on four real datasets and artificial synthetic networks,the rationality and effectiveness of the model are verified.The experimental results show that the algorithm proposed in this paper is slightly better than other algorithms,and more accurate community detection results can be obtained.In the information dissemination model experiment,it is found that the dissemination range expands after considering the strong and weak relationship and community structure,and a more reasonable information dissemination analysis result is obtained.