Real-time event detection based on geo extraction and temporal analysis

Abstract

Microblogging is an important source of information about what is happening in the real world. In this work, we propose a novel approach for realtime event detection targeting accident and disaster events (ADEs) using microblogs from Sina Weibo. Our aim is to detect out every microblog which reports a real-world occurrence of a target event from the microblog stream. We formulate the event detection problem as a classification problem using microblog-based features, linguistic features, content features, and event features. We propose a street-level location extraction method based on the textual content to cooperate geo-information extraction. In order to deliver fresh events, we use a temporal analysis method to filter away past events. We compare our method with two state-of-the-art baselines on event detection, and achieve improvements in both precision and recall.