Analysis of User Influence Using User Behavior and Random Walk

To improve the effect of user influence predicting in microblog, this paper proposed a new method of user influence analysis named UBRWR, based on user interactive behavior and the random walk method. UBRWR firstly used the behaviors between users to reconstruct the network topology, and then an improved PageRank algorithm was applied to predict the user influence, with quantized individual attribute features. The experimental in Weiba show that the UBRWR algorithm outperforms the PageRank algorithm and method using fans count in terms of ranking accuracy.