Topic Extraction from Microblog Posts Using Conversation Structures

Proceedings of the 54th Annual Meeting of the Association for Computational Linguistics, pages 2114–2123, Berlin, Germany, August 7-12, 2016. c⃝2016 Association for Computational Linguistics Topic Extraction from Microblog Posts Using Conversation Structures Jing Li1,2∗, Ming Liao1,2, Wei Gao3, Yulan He4 and Kam-Fai Wong1,2 1The Chinese University of Hong Kong, Shatin, N.T., Hong Kong 2MoE Key Laboratory of High Confidence Software Technologies, China 3Qatar Computing Research Institute, Hamad Bin Khalifa University, Doha, Qatar 4School of Engineering and Applied Science, Aston University, UK {lijing,mliao,kfwong}@se.cuhk.edu.hk1,2 wgao@qf.org.qa3, y.he9@aston.ac.uk4 Abstract Conventional topic models are ineffec- tive for topic extraction from microblog messages since the lack of structure and context among the posts renders poor message-level word co-occurrence pat- terns. In this work, we organize microblog posts as conversation trees based on re- posting and replying relations, which en- rich context information to alleviate data sparseness. Our model generates words according to topic dependencies derived from the conversation structures. In spe- cific, we differentiate messages as leader messages, which initiate key aspects of previously focused topics or shift the focus to different topics, and follower messages that do not introduce any new information but simply echo topics from the messages that they repost or reply. Our model cap- tures the different extents that leader and follower messages may contain the key topical words, thus further enhances the quality of the induced topics. The results of thorough experiments demonstrate the effectiveness of our proposed model. 1 Introduction The increasing popularity of microblog platforms results in a huge volume of user-generated short posts. Automatically modeling topics out of such massive microblog posts can uncover the hid- den semantic structures of the underlying collec- tion and can be useful to downstream applications such as microblog summarization (Harabagiu and Hickl, 2011), user profiling (Weng et al., 2010), event tracking (Lin et al., 2010) and so on. Popular topic models, like Probabilistic La- tent Semantic Analysis (pLSA) (Hofmann, 1999) ∗* Part of this work was conducted when the first author was visiting Aston University. and Latent Dirichlet Allocation (LDA) (Blei et al., 2003b), model the semantic relationships be- tween words based on their co-occurrences in doc- uments. They have demonstrated their success in conventional documents such as news reports and scientific articles, but perform poorly when directly applied to short and colloquial microblog content due to severe sparsity in microblog mes- sages (Wang and McCallum, 2006; Hong and Davison, 2010). A common way to deal with short text sparsity is to aggregate short messages into long pseudo- documents. Most of the studies heuristically aggregate messages based on authorship (Zhao et al., 2011; Hong and Davison, 2010), shared words (Weng et al., 2010), or hashtags (Ram- age et al., 2010; Mehrotra et al., 2013). Some works directly take into account the word re- lations to alleviate document-level word sparse- ness (Yan et al., 2013; Sridhar, 2015). More recently, a self-aggregation-based topic model called SATM (Quan et al., 2015) was proposed to aggregate texts jointly with topic inference. However, we argue that the …

Content truncated to 500 words