Research On Node Influence Mining Algorithm In Complex Networks

Complex network is mainly composed of several nodes and the relationship between nodes.In our real world,relational networks in various fields,including data,physics,biology and computer,have been abstracted and formed many complex networks.Research results show that the network structure formed within the community is common in complex largescale social networks.Community discovery,as the key technology and core problem of complex network research,can achieve further application research by detecting various potential models of complex network.Overlapping community is a core part in the process of community discovery,which is of great significance to the analysis of complex network.In order to solve the problem of high time complexity and poor stability of current overlapping community discovery algorithm,this paper proposes an overlapping community discovery algorithm OCDITN based on three-level neighbor influence analysis in Chapter 3.In this algorithm,TIM is used Calculate the influence of each node,and determine the time order of selecting and updating each node according to the influence of each node;in the strategy of node label updating,determine the updating order of neighbor node label according to the similarity between the node and its neighbors,finally calculate the membership of each node label and find the overlapping community.Compared with SLPA,LPANNI and COPRA,the performance of the proposed algorithm is improved by 7% and 12% on EQ and Qvo,respectively.The result of community partition is more stable and the quality of community partition is higher.In Chapter 4,the traditional label propagation algorithm is extended to the detection of overlapping communities in complex networks,and a label propagation algorithm NILPA based on node influence is proposed.The improved label propagation algorithm can be roughly divided into two main core parts.The first part is the use of LeaderRank.The algorithm calculates the influence of each node,which is the basis of selecting nodes for priority propagation in the second part of the label propagation stage;the second part is the label propagation stage,in which LPA label propagation algorithm is used to process nodes’ community membership,and finally an overlapping community is output.The experimental results on artificial data sets and real data sets show that the reliability of NILPA algorithm is much higher than other comparison algorithms。...