ReTweet ^p: Modeling and Predicting Tweets Spread Using an Extended Susceptible-Infected- Susceptible Epidemic Model

Abstract

Retweeting is one of the most commonly used tools on Twitter. It offers an easy yet powerful way to propagate interesting tweets one has read to his/her followers without auditing. Understanding and predicting tweets’ retweeting extents is valuable and important for a number of tasks such as hot topic detection, personalized message recommendation, fake information prevention, etc. Through the analysis of similarity and difference between epidemic spread and tweets spread, we extend the traditional Susceptible-Infected-Susceptible (SIS) epidemic model as a model of tweets spread, and build a system called ReTweet ^p to predict tweets’ future retweeting trends based on the model. Experiments on Chinese micro-blog Tencent show that the proposed model is superior compared to the traditional prediction methods.