Predicting Information Diffusion Initiated from Multiple Sources in Online Social Networks

Nowadays, online social networks have become a major channel for information dissemination and communication. Many prior studies apply mathematical approaches to characterize and model the process of information diffusion over online social networks. Most of these work focus on the diffusion process of information posted by a single source, however few studies consider the diffusion patterns of information that come from multiple sources. In this paper we first study the basic characteristics of the diffusion process of multi-source informations via real data-sets collected from Digg. Subsequently, we use a mathematical model to predict the information diffusion process of such multi-source news. Finally we validate the accuracy of the proposed mathematical model. Our experiment results show that the model can describe the most representative news stories initiated from multiple sources with an accuracy higher than 90%, and can achieve an average accuracy around 75% across all multisource news stories in the data-set. These results suggest that our approach is able to characterize and predict the spreading patterns of multi-source informations with high accuracy.