User Influence Research Based On Sina Microblog Topic

Social network is gradually changing our lives along with the rise of web2.0, and the form of social network is changing from facebook to renren, from MSN to QQ, from Twitter to sina microblog etc. A variety of social networks are changing the world, not only changed the way of communication between people, but also changed the way of media communication and marketing. As a convenient media interactive platform and efficient means of disseminating information, microblogging has attracted hundreds of millions of users on a global scale, and it has become an important way for people to exchange information and it is increasingly generating extensive and profound impact on the economic and social fields. User is an important part of microblog, and influences and network resources owned by different users directly reflect the huge commercial value and the spread potential. At the same time, the influences of a user in diffenent topics are also different. How to effectively measure user influence in different topics among various users and the comprehensive user influence, and then dig the potential value becomes a problem urgently to be solved.According to the above problem, this paper puts forward a message classification method based on keywords and user influence evaluation method based on improved TwitterRank algorithm. The data set in this paper were from cnpameng.com and the API of sina microblog. The content of the data set is described in detail through the analysis of these data. On this basis, the microblogging messages were classified into topics, and thus lay the foundation for the measure of topics related user influence.TwitterRank algorithm was put forward on the basis of the PageRank algorithm by considering the topic relevance and connection relationship between users to measure the influence of Twitter users. On the basis of TwitterRank algorithm and considering the characteristics of sina microblog, comments, forwarding and release time similarity factors, the TwitterRank algorithm has been improved so that it can accurately messure user influence of sina microblog.The experimental results show that the message classification method based on keywords and the user influence evaluation method based on improved TwitterRank algorithm are correct and effectiveness. During the experiment, Naive Bayes classifier text classification algorithm was used as contrast experiment and the accurancy of message classification method based on keywords was virified. For user influence evaluation method based on improved TwitterRank algorithm, topic related user influence result was analyzed and the relationship between user influence result and friends, followers,message counts and attention were compared. On this basis, overall user influence was analyzed and the conclusion was obtained at last.