Similarity-based emergency event detection in social media

Abstract

Emergency events need early detection, quick response, and accuracy recover. In the era of big data, the use of social media platforms is being popularized. Social media users can be seen as social sensors to monitor real time emergency events. In this paper, a similarity-based method is proposed to early detect all kinds of emergency events in social media, including natural disasters, accidents, public health events and social security events. The method focuses on clustering social media texts based on the 3 W attribute information (What, When, and Where) of events. First, with the two-step classification, emergency related messages are detected and divided into different types from the massive and irrelevant data. Second, the time and location information are respectively extracted with the regular expression matching and the BiLSTM model. Finally, the text similarity is calculated using the type, time and location information, based on which social media texts are clustered into different events. The experiments on Sina Weibo data demonstrate the superiority of the proposed framework. Case studies on some real emergency events show the proposed framework has good performance and high timeliness. As the attribute information of events is extracted during the algorithm flow, it can be described what emergency, and when and where it happened.