Research On Node Influence Ranking Based On Coevolution Propagation In Complex Network

There are many complex relationships in human society.They can be abstracted and simplified as the complex networks,such as friend networks,transportation networks and power transmission networks and so on.Meanwhile,there are many propagating processes in real networks,and such processes can be studied by the spreading models.In the propagating processes,how to select important nodes to promote the information transmission,or how to choose significant nodes to control disease,have always been valuable reasearch issues.Previous researches considered the spread of a single disease or information in complex networks.In real life,there is a high probability that multiple kinds of information or diseases will be coevolution propagation,such as the spread of Spanish flu and pneumonia,and they reinforce each other for cooperative propagation.Therefore,this thesis pays attention to the problem of maximizing the influence of nodes based on cooperative propagation.Then we analyze the influence of target immune strategy based on node centrality on co-infection.This thesis mainly includes the following two aspects:First,we study the problem of maximizing the influence of nodes based on cooperative epidemics.Since two diseases or information in real social networks interact with each other,the spreading dynamics and important nodes metrics are different from single disease that of.In this thesis,we study the problems on the artificial networks and real networks.In cooperative propagation,we select each node to be the initial infected node and get the final outbreak size.We use the outbreak size to sort every node and denote is as true ordering.Then,we use several classical node centralities to sort the nodes and denote them as central ordering.Finally,we find the accuracy function value and the Kendall Tau coefficient to compare the true ordering with central ordering.We find that the network structure is different,and the effect of node centrality is different.Moreover,in the case of local(accuracy function)and global(Kendall Tau),the node centrality is consistent.That is,for a centrality,the smaller of the accuracy function value,the larger of the Kendall Tau coefficient value.Finally,we study the effect of node centrality on spreading model of single disease and multiple diseases.We find that there are many differences in the two cases,which proves the significance of our studies.Second,we consider random immunization and targeted immunization based on central sorting for the cooperative epidemics,and analyze the effects of different immune strategies on co-infection.In the targeted immunization based on central ordering,it is generally believed that the node centrality which accurately predict the important nodes in the network will have a better effect on controlling the disease.However,analyzing the outbreak size of cooperative epidemics and the number of connected clusters of the network after targeted immunization,we find that the targeted immunization based on the betweeness centrality ordering and the closeness centrality ordering work best.However,in some networks,these two centralities do not accurately predict the important nodes.In addition,for random immunization and targeted immunization,the same as traditions,the effect of random immunization is much lower than target immunization.