Design And Implementation Of WeiRank A Voting-model-based Microblog Users’ Influence Evaluation Algorithm

In recent years, various kinds of online social networks developed rapidly. Microblog,as a new kind of online social network, has attract millions of people all over the world.The popularity of microblog not only changes people’s life style, changesthe way of the mass media communication, bring in new e-marketing pattern, but alsodraws academia’s attention on microblog. Microblog has huge amount of users, anddifferent users have different status in microblog network. Meanwhile, there are largeamount of information spreading rapidly on microblog network, which causes theinformation overloading phenomenon. Hence, recommending microblog users byevaluating their influence in microblog network can effectively reduce the informationoverloading in microblog network, and also strongly support the research on online socialnetwork information diffusion and dissemination model. In these respects, a voting modelbased microblog users’ influence evaluation algorithm WeiRank is designed andimplemented. From the view of microblog network topology which consists of users’following relationship, WeiRank algorithm adopts multiple iteration and weighted votingto evaluate the influence score of microblog users.The data for this research comes from Sina Weibo, the popular microblog system inChina. According to the statistics of user’s followers number, followings number as wellas the reciprocal rate, different kinds of users have different influence in the network andthere is a core in Weibo network, which consists of users with strong influence. This coreattracts the majority part of Weibo users. Based on the characteristics of Weibo users’influence, there are two Weibo users’ influence evaluation metrics: one is the coverage ofuser’s influence, the other is influence of the influenced user.After understanding the characteristics of Weibo users’ influence, a voting modelbased microblog users’ influence evaluation algorithm WeiRank is designed andimplemented. WeiRank considering two factors: the coverage of user’s influence andinfluence of the influenced user, assigns two attributes to microblog users: the ability toinfluence other users and ability to identify others. The concept of WeiRank is user’sinfluence score equals to the sum of all his followers’ identify scores, and user’s identifyscore equals to the sum of all his followings’influence scores.User data crawled from Sina Weibo is divided into three different dataset accordingto users’ follower numbers. WeiRank, PageRank and HITS algorithm were run on thesedatasets. The experiment results verified the feasibility and convergence of WeiRank. By comparing with the result of PageRank and HITS, it is proved that WeiRank is moreapplicable to microblog networks.