Research On Node Importance Ranking And Influence Maximization Algorithm In Temporal Networks

With the widespread popularity and rapid development of Internet technology,online social networks have become the main platform for people to communicate and a medium for information dissemination.People’s ability to obtain and transmit information through social platforms such as We Chat,Weibo,and forums has been improved.At the same time,information is spreading rapidly through interactions between users.In social networks,important nodes with greater influence have greater influence on the function and structure of the network.On the one hand,using relevant theories and methods to accurately identify such important nodes in the network is of great significance for accurately placing product advertisements,promoting or inhibiting the spread of information,and suppressing the spread of the epidemic.On the other hand,studying the rules of information dissemination of nodes in the network and guiding the dissemination of information,and finding the set of seed nodes with maximum influence.It is of great significance for realizing the maximization of product promotion and other fields.Taking into account the temporal characteristics of the connections between nodes in social networks,this thesis studies two important topics in the field of important node mining in temporal networks,namely,node importance ranking and influence maximization.At the same time,a temporal network visualization tool was designed and implemented.This thesis mainly carried out the following work:1.Based on the multi-layer graph temporal network model,this thesis proposes an attenuation based supra-adjacency matrix(ASAM)temporal network modeling method based on the attenuation of the coupling strength between layers,and by calculating the eigenvector centrality of the nodes in each time layer network in the temporal network,the importance of the nodes in the temporal network was evaluated.This method also measures the similarity relationship between nodes in adjacent time layer networks and cross-layer networks,and introduces an attenuation factor to more accurately describe the coupling strength between time layers,effectively solving the problem that the fixed constant cannot reflect the difference of the coupling relationship between layers.2.Based on heuristics and greedy strategies,a temporal network influence maximization algorithm is proposed.First,it defines a method for calculating the propagation probability between nodes based on the eigenvector centrality;secondly,it calculates the influence of nodes based on the local information and propagation probability of the nodes in the network;thirdly,the independent cascade model is improved to make it applicable to the information dissemination problem in the temporal network,and on this basis,the temporal network influence maximization algorithm combining greedy and heuristics Strategies(TCHG)was proposed.The experimental results on different temporal datasets show that compared with the existing heuristic algorithm,The TCHG algorithm has a larger range of influence,and the running time is significantly less than the greedy algorithm,which reflects the accuracy and efficiency of the TCHG algorithm.3.Finally,combining the research content of this thesis and the introduction of related theories,a temporal network visualization tool is designed and implemented.This tool maps complex temporal network data and maps it to the GUI interface in the form of a topological graph.At the same time,based on the ASAM temporal network modeling method and the TCHG algorithm proposed in this thesis,the results of node importance ranking and the seed nodes sets with maximum influence are displayed respectively in the GUI interface.Finally,the information dissemination process of the seed node is sequentially displayed based on the independent cascade model.