Message propagation via retweet chain can be regarded as a social contagion process. In this paper, we examine burst patterns in retweet activities. A burst is a large number of retweets of a particular tweet occurring within a certain short time window. The occurring of a burst indicates the original tweet receives abnormally high attentions during the burst period. It will be imperative to characterize burst patterns and develop algorithms to detect and predict bursts. We propose the use of the Cantelli's inequality to identify bursts from retweet sequence data. We conduct a comprehensive empirical analysis of a large microblogging dataset collected from the Sina Weibo and report our observations of burst patterns. Based on our empirical findings, we extract various features from users' profiles, fellowship topology, and message topics and investigate whether and how accurate we can predict bursts using classifiers based on the extracted features. Our empirical study of the Sina Weibo data shows the feasibility of burst prediction using appropriately extracted features and classic classifiers.
展开▼
