In this paper, a novel UR Rank (User Relationships based Ranking) algorithm is proposed for ranking the influence of the user. We first explore five factors that affect user relationship. They are following rate (FR) ...
详细信息
ISBN:
(纸本)9783030009168;9783030009151
In this paper, a novel UR Rank (User Relationships based Ranking) algorithm is proposed for ranking the influence of the user. We first explore five factors that affect user relationship. They are following rate (FR) factor, activity (ACT) factor, authority (ATR) factor, interaction (ITA) factor and similarity (SML) factor. Then those factors are used in supportvectorregression (SVR) model to predict the relationship between users. We assimilate such predicted relationship into a pagerank based transition probability to identify influential users. The experiments on a real micro-blog data set demonstrate that UR Rank algorithm has better performance and is more persuasive than the existing algorithms.
暂无评论