Predicting Micro-blog Individual Retweet Behavior Based on User's Interest Drift

Abstract

Micro-blog post retweet prediction is one of the key technologies for researching information dissemination. The factors that affect retweet are multifaceted; However, the existing methods mainly focus on the research of micro-blog post information attributes and network structure characteristics, and do not fully consider the interests of users that are changing with time. To address this issue, we firstly propose some new methods of measuring features, which are user's influence, user's intimacy with named entities and users' intimacy. Meanwhile, in order to solve this problem of user' s interest changing, we propose a topic model based on user's interest drift to measure user's current interest and calculate topic similarity between users. Finally, we establish a novel micro-blog prediction retweet model based on above features. Our experimental results demonstrate that our proposed method has superior accuracy and stability, compared with several classification methods.