Predicting Popularity of Microblogs in Emerging Disease Event

During emerging disease outbreaks, massive information are disseminated through social network. In China, Sina microblog system as the biggest social network provide a novel way to monitoring the development of emerging disease and public awareness. However, only a small percentage of microblogs could wide spread. Therefore, predict popularity of microblogs timely are meaningful for emergency management. In this paper, a Judgment method for popularity level prediction of microblog is proposed and the temporal pattern between cases number and repost number is verified. Repost number is considered to measure the impact of microblogs. To predict the popularity of microblogs, Granger causality test was used to verify the temporal correlation pattern between development of disease and public concern while an Judgment method based on five classical classification models were proposed. Through analyses, case number of emerging disease are Granger causality of the popularity level of microblogs and the regression model got the best result when lag was three. By Judgment method, more than 86 % microblogs can be classified correctly. The proposed Judgment method based on user, microblog and emerging disease information could analysis the popularity level of microblogs speedily and accurately. This is important and meaningful for monitoring the development of future public health event.