Modeling the Long-Term Post History for Personalized Hashtag Recommendation

Hashtag recommendation aims to recommend hashtags when social media users show the intention to insert a hashtag by typing in the hashtag symbol "#" while writing a microblog. Previous methods usually considered the textual information of the post itself or only fixed-length short-term post history. In this paper, we propose to model the longterm post histories of user with a novel neural memory network called the Adaptive neuralMEmory Network (AMEN). Compared with existing memory networks, AMEN was specially designed to combine both content and hashtag information from historical posts. In addition, AMEN contains a mechanism to deal with out-of-memory situations. Experimental results on a dataset of Twitter demonstrated that the proposed method significantly outperforms the state-of-the-art methods.