Research On Node Detection And Information Propagation In Online Social Network

With the rapid development of Internet and modern technology,online social network becomes a common platform for information communication,opinion expression,commodity marketing and so on.It reduces the cost of information dissemination,broadens the dissemination of information channels,and increases the frequency of public opinion events.Meanwhile,information dissemination becomes intricate and difficult to control because of the uncertainty of online social network topology structure evolution and the diversity of user behavior characteristics.It can not wait to do some research about ascertaining the information dissemination rule of online social network,finding user interactive features,excavating the key transmission users,and setting up effective preventive measures.This thesis mainly focuses on the mechanism of information diffusion in online social network by using interdisciplinary thoughts.The purpose of this thesis is to explore the dynamic genesis and quantify the intensity of the influence,especially provides a theoretical basis for studying the state transition of different user groups in the evolution process.The contribution of this thesis can be summarized as follows:1.As existing epidemic model paid less attention to influence factors and previous research about influence calculation mainly focused on static network topology but ignored individual behavior characteristics,this thesis proposes an information diffusion dynamics model based on dynamic user behaviors and influence.On the one hand,as far as the influence quantification,different from the current research work that mainly focused on network structure,this thesis integrates the internal and external factors,and proposes user influence evaluation method based on the multiple linear regression model.The individual memory principle is analyzed by combining user attributes and individual behavior.User interaction is also studied by using the shortest path method in graph theory.On the other hand,on the modeling of information diffusion,by referring SIR model,the thesis introduces the user influence factor as the parameter of the state change into the epidemic model.The mean-field theory is used to establish the differential equations.2.Since the existing algorithms of node detection mainly considered user relationship but ignored the information interaction behaviors among users,this thesis designs a node decection algorithm named Important Node Rank(INR).Firstly,taking information interaction and links relationship into consideration,the thesis reconstructs the information dissemination network,and uses the entropy method to quantitatively evaluate the user transmission capacity in two dimensions: individual communication will and social communication ability.Secondly,this thesis proposes an INR algorithm based on user propagation capacity by improving the traditional PageRank algorithm.At last,the thesis verifies the proposed model and algorithm by using some data form Tencent Weibo.Experimental results show that the optimized model can effectively combine the topology of information dissemination network and the characteristics of user behavior.This study can not only comprehend the principle and information diffusion mechanism of social influence from a more macroscopic level,but also explain the internal and external dynamics genesis of information diffusion.In addition,the thesis tries to explore the behavioral characteristics and behavior laws of human and improve the accuracy of key user identification.