Research On The Method Of Mining Influential Nodes In Complex Network Based On Topology Structure

Complex networks are pervasive in our daily life,while these networks provide convenience for our lives,certain risk still exists.If a failure occurs in the important nodes,it can cause catastrophic failure of the entire network.Therefore,it is of theoretical and practical significance for identifying influential nodes in complex networks.The centrality of the node is used to measure the importance of nodes in complex networks,this paper is based on the centrality idea,combining the potential importance of the edge and the topology of the network,and this paper studies the ranking of influential nodes in the network and identifies the most influential nodes.The main work is as follows.Firstly,it is analyzed that the correlation between the influence of nodes and degree distribution,clustering coefficient and average path length according to the theory and characteristics of complex networks,and an undirected and unweighted network is applied to analyze the limitations of influence node identification method.Secondly,in the unweighted complex network,in most existing influence node mining algorithm,the problem of the potential importance of the edges and the different contribution degree of the neighbors is not considered,a novel influence node algorithm based on Two-Degree Centrality TDC is proposed in this paper.The potential important of edge is defined,by considering the differentiated contribution degree of its neighbors,and the tuning parameter ? is introducing to adjust the influence degree of different neighbors.Thus,the influence value of each node is obtained in the network.Thirdly,in the weighted complex network,in most existing influence node mining algorithm,the problem of the network topology of nodes and the contribution degree of different neighbors is not considered,a novel influence node algorithm based on Clustering Degree Algorithm CDA is proposed in this paper.On the basis of the the weighted degree,the network topology is further considered,and consideration is given that the differentiated contribution degree of direct neighbors,each neighbor's contribution to the corresponding node is measured by the edge weight.Thus,the influence value of each node is obtained,these the influence values are sorted by descending,the higher the value is,and the greater the influence of the node is.Finally,the experiments are done on the platform of Windows,and the MATLAB language is used.Different real networks are applied for the effectiveness comparison between the proposed algorithms and other algorithms from different aspects.