Research On Node Influence Measurement And Influence Maximization Of Complex Network

Network influence analysis is a hot topic in complex network research,which includes the measurement and maximization of node influence,the former is to evaluate the influence of each node in the network,the latter is to evaluate the overall influence of the node set in the network,and select the set with the largest spread range These two studies are of great practical significance for bioinformatics and genomics,marketing,infectious disease control and network monitoring.Based on the research status and existing problems of node influence measurement and influence maximization,a new influence measurement method is proposed,and based on which three kinds of influence maximization algorithms are further put forward.The specific contents are summarized as follows:Above all,because K-shell is a coarse-grained measurement method,it can't identify the nodes in the key position effectively,a node influence measurement method KSSH is proposed,which integrates k-shell method and structural hole index.The correlation and resolution indexes show that KSSH method has good accuracy and discrimination.In addition,KSSH method is introduced into the influence maximization problem,and the influence overlap effect between nodes in the network is discussed.This paper uses KSSH method to evaluate the influence of each node,and based on the clustering coefficient of node,the number of selected seed nodes in neighbor nodes,the propagation probability and the first-order and second-order neighborhood of nodes,two influence maximization algorithms KSSH_DisN and KSSH_DisN+ are proposed to reduce the influence of overlapping.The comparative experiments of multiple data sets in Independent Cascade Models prove that the two algorithms can achieve similar influence with the CELF algorithm.What's more,in order to further improve the accuracy of the algorithm,a hybrid influence maximization algorithm CKS_CELF based on the candidate set is proposed.This algorithm divides the solution process into heuristic phase and greedy phase.The heuristic phase adopts KSSH method to evaluate the influence of nodes,and a small part of nodes with large influence is selected as candidate set from each community by using the community structure of the network,and then it selects k-seed nodes from the candidate set by CELF algorithm.The experimental results show that CKS_CELF algorithm can not only achieve the same or even slightly higher accuracy with CELF algorithm,but also shorten the running time of the algorithm.