Ranking User Tags in Micro-Blogging Website

Abstract

Users can annotate themselves using free tags in micro-blogging website such as Sina Weibo. The tags of a user demonstrate the characteristics of the user and are generally in a random order without any importance or relevance information. It limits the effectiveness of user tags in system recommendation and other applications. In this paper, we proposed a user tag ranking schema which is based on interactive relations between users. Influence strength between users is considered in our user tag ranking method. Relevance scores between tags and users are also utilized to rank user tags. Experiments are conducted on distributed processing framework Hadoop to process the very large Sina Weibo dataset which contains more than 140 million users. Experimental results show that our method outputs frequently used method and gives good performance.