Research On Dynamic Community Detection Algorithm Of Online Social Networks

Since complex networks can be used to model most complex scenes and systems,the research on complex networks has received extensive attention and exploration from all walks of life,and the theory and application of complex networks have also made great progress and development.As one of the main features of complex networks,community structure is the main focus in the field of complex network research.Although the current research on static complex network community detection has proposed many effective methods,as the research deepens,researchers find that dynamic characteristics exist in most actual networks.With the passage of time,the nodes and the edges in the networks will change accordingly,and the community structure and its evolution law in the dynamic networks can be found,which can better analyze and observe the hidden contents,hierarchical relationships and topologies in the networks.Therefore,the research on dynamic networks community detection algorithms has more important significance and value.This paper first expounds the basic concepts of complex networks and community detection,then analyzes several commonly used static community detection algorithms and dynamic community detection algorithms,and summarizes the advantages and disadvantages of current community detection algorithms.In view of the community partitioning results,partitioning accuracy and excessive artificial thresholds in the Evolutionary Community Structure Discovery algorithm(ECSD)in dynamic weighted networks,this paper proposes an algorithm that combines the structure of historical network communities,called the Evolutionary Community Detection Algorithm(ECDA)in dynamic weighted networks.This algorithm is divided into two steps:(1)taking the historical community structure information and the historical network structure information as the reference,synthesizing the current timestep network structure,calculating the current time step input matrix,(2)then,the community partitioning result combining historical timestep information can be obtained through the calculation of input matrix.In this paper,the algorithm is verified experimentally on artificial network datasets and real network datasets.The experimental results show that the proposed algorithm in this paper has the following advantages:(1)It can automatically discover the community structure of each timestep in the dynamic weighted networks,(2)this algorithm has a higher sensitivity to changes in network structure and changes in community structure.Compared with other classical algorithms,the proposed algorithm can effectively discover the community structure in dynamic weighted networks,and has better stability,robustness and competitiveness.Moreover,this algorithm only needs simple network node relationship data,and does not need network information such as the number of communities,node tags,etc.,and the datasets are convenient to obtain,thereby having higher application value and greater application range.