Discovery Of Influence Of Network Spread Node Based On Betweeness Centrality

With the popularity of the Internet,the Internet has gradually penetrated into all aspects of life.Instant messaging,such as QQ,WeChat,in the early days of the Internet has been a very good development.In today's era of popular Internet,SNS micro blog,such as Twitter,Facebook and Weibo,gradually make the relationship of betweeness people become more direct and more complex,the flow of information has become more rapidly and immediately.Through research and analysis,to find the most influential Top-K nodes in social networks in the massive social network data,which plays a decisive role in the dissemination of information in the social network.Influence Top-K node mining algorithm has become the hot spots for people to study and improve.This paper is mainly based on the Betweeness Centrality algorithm,aiming at improving the efficiency and accuracy of the Betweeness Centrality algorithm.Firstly,the random walk data sampling is carried out on the massive social network data.Secondly,the improvement of Betweeness Centrality algorithm based on different and local factor is adopted on the extracted samples.In the past,most of the research objects of social network influence are non-right undirected graphs.In order to close to reality,the subjects of this paper is directed graph.?1?Due to the rapid development of the Internet,it provides a huge amount of data for social network research.However,if the amount of data is too large,it will influence the influence research.For this problem,this paper improves the existing random walk sampling algorithm,and proposes the random walk data sampling algorithm based on unequal probability of directed graph?DUPRW?.Excellent sampling algorithm can accurately reflect the characteristics of the overall set,but it is difficult for the traditional sampling method to characterize the complexity,uncertainty and so on of the social network.This paper first divides the community into social networks and adds virtual neighbor nodes to each node.Then randomly selects the nodes in the community as the starting node Vi,marks its neighbor nodes and selects the neighbors randomly or jumps on the network with the probability?.This step is repeated as the next step of the random walk.When the random walk is trapped in the local subgraph,the nodes with larger degrees in the unsampled set are selected as the starting node of the next random walk sampling,which base on unequal probability.After the sampling is completed,the remaining nodes are sampled to select the node degree is greater than a threshold and the inter-community connection nodes as supplementary nodes to join the sample to avoid the important nodes are not sampled,which can accurately describe the complex structure of the social network.The extracted sample will be used as the algorithmic data set for the social network influence Top-K node discovery.?2?First,the traditional Betweeness Centrality algorithm treats the shortest path through the node V_i and the position of the node in the shortest path as the same effect.In this paper,based on the attenuation model of information in the propagation of the shortest path of different lengths do different treatment,the information is transmitted once,the amount of information attenuation before the??0<?<1?times.Secondly,the traditional Betweeness Centrality algorithm has a very accurate characterization of the importance of node V in the whole social network,but ignores the influence of the node V in the local network.In this paper,by adding the local influence of node V,as a complement to the overall influence of node V,and the different Betweeness Centrality algorithm based on local factors?DLBC?is proposed.Finally,through the experiments of correlation in different datasets,the validity of the Different Betweeness Centrality algorithm based on local factors is proved.