Predicting Information Popularity Degree in Microblogging Diffusion Networks

Microblogs have rapidly become the most popular means by which people communicate with friends, pay close attention to celebrity at any time. Hence many studies on microblogging networks have been done recently, focusing on information diffusion, popularity prediction, topic detection and more. In this paper, we study the popularity of tweets in microblogging networks and introduce a novel concept “popularity degree ” that help divide microblogging into four levels. Through the empirical analysis of different popularity degree, we find the retweeting information of a tweet at an earlier time can help predict its final popularity. Hence we propose a prediction model based on SVM with the retweeting information within one hour. Experimental results show our model has better ability of prediction.