Automatic Hashtag Recommendation for Microblogs using Topic-Specific Translation Model

Abstract

Microblogging services continue to grow in popularity, users publish massive instant messages every day through them. Many tweets are marked with hashtags, which usually represent groups or topics of tweets. Hashtags may provide valuable information for lots of applications, such as retrieval, opinion mining, classification, and so on. However, since hashtags should be manually annotated, only 14.6% tweets contain them (Wang et al., 2011). In this paper, we adopt topic-specific translation model(TSTM) to suggest hashtags for microblogs. It combines the advantages of both topic model and translation model. Experimental result on dataset crawled from real world microblogging service demonstrates that the proposed method can outperform some state-of-the-art methods.