Identifying Event-related Bursts via Social Media Activities

Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pages 1466–1477, Jeju Island, Korea, 12–14 July 2012. c⃝2012 Association for Computational Linguistics Identifying Event-related Bursts via Social Media Activities Wayne Xin Zhao†, Baihan Shu†, Jing Jiang‡, Yang Song†, Hongfei Yan†∗ and Xiaoming Li† †School of Electronics Engineering and Computer Science, Peking University ‡School of Information Systems, Singapore Management University {batmanfly,baihan.shu,yhf1029}@gmail.com,ysong@pku.edu.cn jingjiang@smu.edu.sg, lxm@pku.edu.cn Abstract Activities on social media increase at a dra- matic rate. When an external event happens, there is a surge in the degree of activities re- lated to the event. These activities may be temporally correlated with one another, but they may also capture different aspects of an event and therefore exhibit different bursty patterns. In this paper, we propose to iden- tify event-related bursts via social media activ- ities. We study how to correlate multiple types of activities to derive a global bursty pattern. To model smoothness of one state sequence, we propose a novel function which can cap- ture the state context. The experiments on a large Twitter dataset shows our methods are very effective. 1 Introduction Online social networks (e.g., Twitter, Facebook, Myspace) significantly influence the way we live. Activities on social media increase at a dramatic rate. Millions of users engage in a diverse range of routine activities on social media such as posting blog messages, images, videos or status messages, as well as interacting with items generated by oth- ers such as forwarding messages. When an event interesting to a certain group of individuals takes place, there is usually a surge in the degree of ac- tivities related to the event (e.g., a sudden explosion of tweets). Since social media activities may indi- cate the happenings of external events, can we lever- age on the rich social media activities to help iden- tify meaningful external events? This is the research problem we study in this paper. By external events, we refer to real-world events that happen external to the online space. ∗Corresponding author. 2 4 6 8 10 0 20 40 60 80 100 120 140 Time index all�tweets retweets url�embedded tweets Noise (a) Query=“Amazon.” 2 4 6 8 10 0 50 100 150 Time index all�tweets retweets url�embedded tweets Noise (b) Query=“Eclipse”. Figure 1: The amount of activities within a 10-hour window for two queries. Three types of activities are considered: (1) posting a tweet (upward triangle), (2) retweet (downward triangle), (3) posting a URL- embedded tweet (excluding retweet) (filled circle). As explained in Table 1, both bursts above are noisy. Mining events from text streams is usually achieved by detecting bursty patterns (Swan and Al- lan, 2000; Kleinberg, 2003; Fung et al., 2005). How- ever, previous work has mostly focused on tradi- tional text streams such as scientific publications and news articles. There is still a lack of systematic in- vestigations into the problem of identifying event- related bursty patterns via social media activities. There are at least two basic characteristics of …

Content truncated to 500 words