Average Edge-distance Contribution Algorithm For Identifying The Influential Nodes In Complex Networks

How to accurately and effectively identify the influential nodes in networks has always been a core issue.In order to evaluate the importance of the nodes,some famous centrality algorithms have been proposed and widely used.However,the existing algorithms still remain some shortcomings.In order to effectively identify the influential nodes in the network,this paper utilizes the concept of multi index comprehensive analysis in statistics and proposes a recognition algorithm based on entropy weight-gray correlation method(EG).By constructing the index data matrix and using the weighted gray correlation analysis method to calculate the corresponding correlation degree of each node and reflect the influence of nodes.The results show that it is feasible to use the index comprehensive analysis to evaluate the influence of nodes.But we found that the weight of the betweenness centrality is relatively large,in the four networks,the ability of identifying the influence of the nodes is weak for the betweenness centrality method.The EG algorithm needs to fully integrate the index data information,so the betweenness centrality with the large weight will inevitably affect the efficiency of the EG algorithm.In order to avoid the adverse effect of index data on algorithms,in this paper,a novel method named average edge-distance contribution(AEDC)that is not completely dependent on index data is proposed,which measures the average contribution of each edge to the sum of distances of all node pairs in the network.For each node,we utilize the relative change of AEDC by removing it from the network to determine its influence.For verifying the effectiveness and feasibility of the AEDC method,we simulate the process of disease spreading in four real complex networks with the Susceptible-Infected-Recovered(SIR)model.The experimental results show that our proposed method is more accurate than several benchmark centrality measures in terms of identifying the influential nodes.