Research On Influence Maximization Based On Path Analysis In Complex Networks

With the rapid development of computer technology and the rise of social media in recent years,some social media(such as Facebook,Weibo,Linkedln,WeChat,etc.)have gradually become a platform for people to disseminate and share information.Compared with traditional mass media,information dissemination in social media has the advantages of high efficiency and high credibility,but at the same time,the spread of inappropriate lyrics will cause great harm to social stability.Therefore,this thesis studies the influence maximization based on complex networks,focusing on how to maximize the influence in the network based on the method of propagation path-analysis.The specific research contents are as follows:(1)We propose a maximum likelihood-based algorithm named MLIM for the influence maximization problem under the independent cascade model.First,we constructed a thumbnail of the social networks based on sampling scheme.Then,we pursued maximum likelihood to select the top-k influential seed nodes based on the thumbnail.Experiments on real-world networks show that compared with the traditional Greedy algorithm,the MLIM algorithm performs better and has lower time-consumption.(2)We propose a propagation path analysis-based algorithm named GREEDY-SIM to solve the influence maximization in the signed networks.First,we construct an independent cascade model in signed networks,and find the first m shortest paths between each node pair by the single source shortest path method.Then we calculate the number of positive influence nodes and negative influence nodes,and finally select the k nodes with positive states as the seed node set.The experimental results on real-world networks have demonstrated that the proposed method not only can obtain the widely influence spreading,but also be more accurately.(3)We propose an efficient influence maximization algorithm IMPP with limited unwanted users(IML)in competitive online social networks.The first step is to find all independent paths L between the nodes v and u,and then use the independent path L to calculate the activation probability value a(v,u).The second step analyzes the independent path of the candidate nodes and calculates the propagation increment ?s(x)after joining a node.Finally,the seed set S is updated according to the change of ?s(x).The key point of this method is to determine the nodes that hinder the influence propagation when determining the seed nodes.The experimental results on real-world networks have proved that the IMPP algorithm can achieve better influence propagation at a lower time consumption.