Research And Application Of Community Discovery And Evolution In Social Networks

With the advent of the social era,social network analysis has gradually become a research hotspot,among the many problems in social network research,community discovery and evolutionary analysis is a typical problem.Compared with other complex networks,social networks are more complex in terms of node diversity and structural dynamics.Discovering and analyzing the process of community evolution in social networks can help deepen your understanding of social networks at a meso level,at the same time,the results of community analysis are of great help to research in other fields.The research content of this article includes the following aspects:(1)Aiming at the problem that the existing static community discovery algorithm has poor stability in the process of seed node selection and is easily affected by the processing order during expansion,resulting in poor accuracy of the division result,a overlapping community discovery algorithm based on Node2 Vec.First,use the Node2 Vec algorithm to obtain a vector representation of each node in the network and calculate the similarity between any two nodes.Then,use the improved node influence function of this paper to calculate the influence value of each node,select the node with the largest local influence value as the seed node,and divide the node to be updated into the corresponding community according to the similarity between the nodes until all nodes belong to at least one community.Experiments show that the algorithm can more accurately mine the community structure in social networks under the premise of ensuring stability.(2)Aiming at the problems of the existing incremental dynamic community discovery algorithm that is prone to error information accumulation and the community discovery results are susceptible to the incremental sequence,an incremental dynamic community discovery algorithm based on seed node reselection and expansion is proposed.Mark all the increment itself and the nodes that may be affected as the status to be updated,reduce the possibility of error information accumulation by expanding the scope of the increment,and compare the similarity between the node to be updated and each seed node.The size relationship of the threshold value,the nodes within the update range are divided independently to avoid the influence of the incremental order on the experimental results.Experiments show that the algorithm proposed in this paper shows a higher level of stability and accuracy on both artificial and real datasets.(3)Combining the idea of seed nodes with the process of community evolution analysis,a community tracking strategy based on the comparison of seed nodes is proposed.By comparing the relationship of seed nodes between networks at adjacent moments,the independent community evolution sequences are separated out.In the sequence of community evolution,it independently analyzes the occurrence of various community evolution events,then summarizes the information,and then conducts related analysis of various evolution events in combination with network background information.