Topic Model for Identifying Suicidal Ideation in Chinese Microblog

Topic Model for Identifying Suicidal Ideation in Chinese Microblog Xiaolei Huang1, Xin Li1, Lei Zhang2, Tianli Liu3, David Chiu4, Tingshao Zhu*5,6 1Computer Network Information Center, Chinese Academy of Sciences (CAS), China, {xlhuang,lixin}@cashq.ac.cn 2Jinan University, China, zl.ramiel@gmail.com 3Inst. of Population Research, Peking University, China, tianli.liu@pku.edu.cn 4Dept. of Math and Computer Science, University of Puget Sound, USA, dchiu@pugetsound.edu 5Inst. of Psychology, CAS China, tszhu@psych.ac.cn 6Inst. of Computing Technology, CAS China Abstract Suicide is one of major public health problems worldwide. Traditionally, suicidal ideation is assessed by surveys or interviews, which lacks of a real-time assessment of personal men- tal state. Online social networks, with large amount of user-generated data, offer opportu- nities to gain insights of suicide assessment and prevention. In this paper, we explore potentiality to identify and monitor suicide expressed in microblog on social networks. First, we identify users who have commit- ted suicide and collect millions of microblogs from social networks. Second, we build suicide psychological lexicon by psychologi- cal standards and word embedding technique. Third, by leveraging both language styles and online behaviors, we employ Topic Model and other machine learning algorithms to iden- tify suicidal ideation. Our approach achieves the best results on topic-500, yielding F1 − measure of 80.0%, Precision of 87.1%, Re- call of 73.9%, and Accuracy of 93.2%. Fur- thermore, a prototype system for monitoring suicidal ideation on several social networks is deployed. 1 Introduction Suicide is a severe health problem worldwide, which is one of leading causes of youth death in the world, especially in China. In the latest report (Organiza- tion and others, 2014) from World Health Organi- zation (WHO), over 800, 000 people committed sui- cide in 2012, including 120, 730 Chinese; and it is very likely that the data is underestimated. In- deed, there are many more people who attempt sui- cide every year. Instead of calling health services or seeking for help in-person, choosing social networks is a preferable choice for some suicide because of privacy and facilitating sharing similar experiences among peers (Luxton et al., 2011). Social network sites (SNS), such as Twitter, Sina Weibo, have become popular platforms for peo- ple to express themselves. Sina Weibo is a Chi- nese leading social network akin to Twitter. Ac- cording to the latest Sina Weibo User Activity Report (http://data.weibo.com/report/ reportDetail?id=215), Weibo now has more than 70 million active users per day, and over 160 million active users per month. It becomes a great platform for sharing opinions, emotions, and even to breaking news or public events. Recent work (Fu et al., 2013) showed that SNS not only enhanced our connections with others, but also facilitated selec- tive self-presentation of undesirable behaviors, such as suicide. The association between social media and suicide has drawn public attention recently, since several actual suicidal cases were reported in Sina Weibo, e.g., (http://news.sina.com.cn/zl/ zatan/2014-12-02/18032759.shtml). However, new approaches towards online suicide ideation monitoring and prevention are still under development. (Fu et al., 2013) suggested that dif- fusion of microblogs about one’s suicidal ideation or behaviors …

Content truncated to 500 words