Context-Aware Topic Modeling for Content Tracking in Social Media

Content in social media is difficult to analyse because of its short and informal feature. Fortunately, some social media data like tweets have rich hashtags information, which can help identify meaningful topic information. More importantly, hashtags can express the context information of a tweet better. To enhance the significant effect of hashtags via topic variables, this paper, we propose a context-aware topic model to detect and track the evolution of content in social media by integrating hashtag and time information named hashtag-supervised Topic over Time (hsToT). In hsToT, a document is generated jointly by the existing words and hashtags (the hashtags are treated as topic indicators of the tweet). Experiments on real data show that hsToT capture hashtags distribution over topics and topic changes over time simultaneously. The model can detect the crucial information and track the meaningful content and topics successfully.