Research On User Influence Algorithms Based On Microblog Community

As a widely-used information carrier,microblog not only changes the users' communicative habits as well as leads to the rise of microblog marketing and product recommendation,but also attracts the attention and research of the academic community.In recent years.It has become one of the most popular social networks.This thesis,focusing on the characteristics of ignoring the single microblog value,the content of the blog and basing on a comprehensive consideration of the users' behavioral attributes and microblog content,proposes an evaluation model of microblog influence and solves the problems the users have in the influence evaluation.Firstly,in the current research on the evaluation model of microblog influence,neglecting the value of single microblog information generally focusing on the influence of users and the number of the user's behavioral attribute forwarding,comments as well as thumbing up is truly the recognition of the value of microblog information.The differences of regional economic conditions and the popularity of the Internet will lead to different influences of the same number of microblog forwarding,comments,and thumbing up in the network.Therefore,the thesis adds the concept of coverage and influence saturation and conducts the study of the value of microblog information dissemination from these perspectives.Secondly,with the problem of focusing on the users' behavioral attributes in the current influence measurement algorithm,ignoring microblog contents,the value of the node itself as well as users' interest preferences etc,this thesis,embarking on the users' information and basing the linear weighted model,makes a comprehensive analysis of users' behavioral attributes,blog similarity as well as node similarity and establishes an index system of users' influence evaluation.Thirdly,utilizing the index system of user influence evaluation created in(2),the author proposes User Influence Measurement Rank(UMR)based on users' behaviors and blog contents.This algorithm uses the Page Rank thought to evaluate the influence of microblog.Given the deficiency of Page Rank's not considering the difference of user nodes and the unevenly-distributed PR value,the UMR algorithm analyzes the behavioral attribute of nodes and the topic preferences between users and includes the content of the blog as well as the similarity of the structure,thus making the nodes have certain property preferences and topic similarity in the process of assigning the PR value and improving the comprehensiveness,veracity and objectivity of the algorithm.By using the UMR results of influence ranking are compared with PR algorithm and the BRR(Behavior-Relationship Rank)algorithm.it is verified that the UMR algorithm can objectively reflect the interest preference of interactive behaviors between users based on the content of the blog,thus eliminating the influence of zombies users on the ranking.With the rapid development of microblog,it is becoming more and more important to tap the potential and high-impact users.The microblog influence evaluation model proposed in this thesis provides a basis for further research and extension of microblog.