Microblog oriented interest extraction with both content and network structure

Abstract

Microblog is an important social media platform, which is a microcosm of microblog users’ real life, which it possible to obtain user’s real intention and interests by identifying user interest from microblog. In the literature, most existing approaches for extracting user interests usually make use of only the information contained either in the textual posts, or in the social network structure of microblog. In this paper, we propose a systematic framework for interest extraction taking both the textual and social network information of microblog into account to get high quality tags. We first extract users’ candidate interests based on the content of microblog, then propose a graph-based approach UNITE based on social network information for ranking user interest, finally introduce a more reasonable and objective metric for evaluation. Experimental results on Sina Weibo, one of the most popular microblog in China, demonstrate that our proposed approach makes dramatic improvements over state-of-the-art baselines.