Social Network Immunization Via Higher-order Organization

The rapid development of information technology and online social network platform not only shortens the distance of the world and greatly improves social efficiency,but also brings great hidden dangers and challenges.At present,the flooding of bad information on the network has become a global problem.A large number of harmful information,with the help of increasingly large and powerful online social network media and applications,can spread rapidly in a very short time in a country or even in the world,causing serious misleading and cognitive bias to a large number of people in society,and even irrational phenomena to groups,resulting in serious consequences.Therefore,immunization of social networks has attracted increasing attention over the last decade.Various algorithms have been proposed based on the topo-logical structure of networks,such as the degree and betweenness of nodes.However,most of these studies have only observed the basic topological structure at the level of individual nodes,ignoring higher-order structures captured by network motifs,which may lead to insufficient performance.In addition,immunization based on the connectivity pattern of nodes such as the degree in a social network may cause integrity problems and also interfere in other users' regular activities because the absence of the hub nodes can greatly impair the connectivity of the network.Thus,to settle the issues mentioned above,following works have been done:First,in this paper,the high-order implicit structure of the network based on the motifs is studied.By studying the connectivity of motifs and the weight graph of the network based on the motifs,a method of extracting the high-order connectivity of the network through the adjacency matrix based on the motifs is obtained.Second,in this paper,the prominent role of edge-betweenness centrality in network immunization is studied.It is found that this metric can not only make the proposed algorithm more efficient than the traditional algor:ithm,but also play a more outstanding role in reducing its destructive effect on network structure and function while the immunization.Third,combining the high-order structure of network based on motifs and the edge-betweenness centrality,a new importance index of edges called MEBC(Motif-based Edge-Betweenness Centrality)for network immunization is proposed in this paper,and a network immunity algorithm with the same name is proposed on the basis of MEBC.Finally,in the experiments,the SIR model is used to simulate the information transmission process in the network.The proposed MEBC algorithm is compared with other algorithms to simulate and test its immune performance on real social network data sets.A large number of experimental results show that the proposed immune algorithm MEBC not only has good immune performance,but also has better impact on network integrity than other algorithms.