Twitter Catches The Flu: Detecting Influenza Epidemics using Twitter

Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, pages 1568–1576, Edinburgh, Scotland, UK, July 27–31, 2011. c⃝2011 Association for Computational Linguistics Twitter Catches The Flu: Detecting Influenza Epidemics using Twitter Eiji ARAMAKI Sachiko MASKAWA Mizuki MORITA The University of Tokyo The University of Tokyo National Institute of JST PRESTO Biomedical Innovation Tokyo, Japan Tokyo, Japan Osaka, Japan eiji.aramaki@gmail.com sachiko.maskawa@gmail.com morita.mizuki@gmail.com Abstract With the recent rise in popularity and scale of social media, a growing need exists for systems that can extract useful information from huge amounts of data. We address the issue of detecting influenza epidemics. First, the proposed system extracts influen- za related tweets using Twitter API. Then, only tweets that mention actual influenza patients are extracted by the support vector machine (SVM) based classifier. The ex- periment results demonstrate the feasibility of the proposed approach (0.89 correlation to the gold standard). Especially at the out- break and early spread (early epidemic stage), the proposed method shows high correlation (0.97 correlation), which out- performs the state-of-the-art methods. This paper describes that Twitter texts reflect the real world, and that NLP techniques can be applied to extract only tweets that contain useful information. 1 Introduction Twitter1, a popular micro-blogging service, has received much attention recently. It is an online network used by millions of people around the world to stay connected to their friends, family members, and co-workers through their computers and mobile telephones (Milstein et al., 2010). Nowadays, Twitter users have increased rapidly. Its community estimated as 120 million worldwide, 1 http://twitter.com/ posts more than 5.5 million messages (tweets) eve- ry day (reported by Twitter.com in March 2011). Twitter can potentially serve as a valuable infor- mation resource for various applications. Huber- man et al. (2009) analyzed the relations among friends. Boyd et al. (2010) investigated commuta- tion activity. Sakaki et al. (2010) addressed the detection of earthquakes. Among the numerous potential applications, this study addresses the is- sue of detecting influenza epidemics, which pre- sents two outstanding advantages over current methods.  Large Scale: More than a thousand messages include the word “influenza” each day (Nov. 2008 – Oct. 2009). Such a huge data volume dwarfs traditional surveillance resources.  Real-time: Twitter enables real-time and di- rect surveillance. This characteristic is ex- tremely suitable for influenza epidemic detection because early stage detection is im- portant for influenza warnings. Although Twitter based influenza warnings poten- tially offer the advantages noted above, it might also expose inaccurate or biased information from tweets like the following (brackets [] indicate the comments):  Headache? You might have flu. [Suspi-­‐ cions]  The World Health Organization reports the avian influenza, or bird flu, epidemic has spread to nine Asian countries in the past few weeks. [General News] 1568  Are you coming down with influenza? [Question] Although these tweets include mention of “influ- enza” or “flu”, they do not indicate that an influen- za patient is present nearby. We regard such messages (merely suspicions/questions, general news, etc.) as negative …

Content truncated to 500 words