科研之友
微信
新浪微博
Facebook
分享链接
摘要
Similar to Twitter, Weibo is a popular Chinese microblogging service that is used to read and write millions of short text messages on any topic within 140-character limit. Users create status messages, which sometimes show opinions about different subjects. Particularly, after a disaster, people always express their states and emotions toward the situation via microblogging service. The previous study works revealed that public negative emotions could be associated with the subsequent incidents. Therefore, once a disaster happens, the crowed negative sentiment among victims needs to be paid more attention, which could be useful to discover the following emergency events such as public fear and crisis. In order to detect potential incidents implicated by victims' negative emotions in the post-disaster situation, this paper proposes a structured framework including three phases. The first phase focuses on how to identify disaster-related Weibo messages from the massive and noisy microblogging stream, and the second phase is about how to filter negative sentiment messages from all of the disaster-concerned microblogging. We introduced machine learning methods into both of the above phases. In the last phase, we pay attention on crowd negative sentiment, by tracking and predicting victims' negative emotions changing trend on the base of GM (1, 1) to carry out incidents discovery in a post-disaster situation. By the case study of Ya'an earthquake, we demonstrated that the proposed framework could perform well in incidents monitors such as aftershocks and potential public crisis, which is meaningful and useful to disaster relief process and emergency management in post-disaster situation.
- 出版日期2016-9
- 单位哈尔滨工业大学
