Identifying Influential Users by Improving LeaderRank

Abstract

In large-scale social network, influential users play important roles in public opinion analysis and information promotion. So, identifying influential users attract increasing attention in social network studies, and researchers have been trying to find a better algorithm to identify influential users. PageRank is useful for searching information in World Wide Web, but it has less effectiveness in social networks. LeaderRank is improved by PageRank and it has better ranking effectiveness than PageRank. But LeaderRank doesn’t consider the impact of the various personal attributes of the users. Therefore, this thesis proposes a new algorithm, the ANiceRank, which combines with the users’ multiple properties. To test the effectiveness of our algorithms, we introduce SIR model, which is widely used in virus transmission and information dissemination in social networks. The results show that the accuracy of the improved algorithm has been effectively improved and it can be used to identify influential users in social networks.