Survey on Predicting Popularity of Information in Microblogs

Microblogs have changed the traditional way of information dissemination. This paper presents a survey of predicting information popularity in microblogs. Based on the definition of the popularity prediction problem of microblogs, the main factors which affect the popularity of microblogs are introduced from the aspects of content, user information and network structure. Four methods based on time series, epidemic model, classification model and regression model are compared. The future direction of the popularity prediction is pointed, which is focused on the time granularity, minimum observation time and dynamic prediction of popularity. This study is important to information retrieval, public opinion monitoring, business marketing and so on.