Research And Evaluation On Influence Of Microblog User Based On Interactive Behavior

Microblog as the newly emerging web2.0era of information dissemination andsharing of social network service platform, for the high openness and interactivityadvantages such as strong, powerful information dissemination ability, has become animportant kind of information carrier and transmission medium. In many social hotissues discussed, the outbreak of the network public opinion, government corruption,network marketing, etc, we can see microblog. In our daily work, life and even thesociety, microblog has produced important influence. Microblog contains hugeinfluencewhichhasmadeitgraduallybecomethefocusofscholarsathomeandabroad,also attracted various fields of scientific research personnel involved in the study ofmicroblog. But as microblog users growing, microblog message swelling, an inevitableproblem occurs, that is information overload reducing identification of people.Therefore, to analyze the influence of the users in microblog, excavated microblog mayplay a key role of information transmission in the social network, which has a greatinfluence, is particularly important.In this paper, the influence of microblog users are defined as the ability the userhas to influence other people’s thoughts or inspire others to generate more interactivebehavior through the dissemination of information or interaction with others. Based onthe definition of influence, this paper proposed a new microblog user influenceevaluation model to solve the problems existing in the original Pagerank algorithm,such as the original Pagerank distributes edge weight averagely, and less susceptible to"zombie fans". The evaluation method considered the microblog network structure, theinteractions between the users and the user’s own activity to make the evaluationreasonable.Around the research topic of this article, this paper has done following jobs:First of all, based on microblogin the analysis of the interaction between the users,and combined with the propagation characteristics of microblog, microblog userinteraction model is established. Through the analysis of the four elements ofinteraction model, this paper determines the three key factors, that rate of message was received, user’s activity, the interaction strength between the receiver andcommunicators. Combining these three factors, we defined the influence transfercoefficient between the users for user influence evaluation index.Then，add the influence transfer coefficient to Pagerank algorithm as the weighttransfering from user to user. The MUR-IBM (Microblog User Rank-based onInteractive behaviors Model) algorithm was proposed. Based on GraphChi framework,we developed the MUR-IBM algorithm with C++language. This algorithm can dograph calculation of millions edges in reasonable time.Finally, we used the API provided by Tencent Weibo to get microblog user data.With these data, we used MUR-IBM to calculate every user’s influence, and gave outthe rank list. Then we find that MUR-IBM can relect the interactions between usersmore effectively, and reduce the affect by inactive "Zombie fans". This verifies thevalidity of the model.