A novel information diffusion model for online social networks

Abstract

Online social networks form a central platform for information sharing and influence maximization. Even though the information dissemination within online social networks flows naturally as diffusion process, the dynamics of online social networks make it challenging to model precisely the spreading mechanism and enable concise prediction of information diffusion. In this paper, we propose information diffusion model in online social networks that applies the physical radiation energy transfer phenomena. To our knowledge, this model is the first in domain that adapts the natural radiation transfer as predictive information diffusion model. Physically, the radiation transfers from one surface to another which has similar form of diffusion in online social networks where the contagion transfers from one user to another. Theoretically, the proposed RADiation DIFFusion (RADDIFF) model measures and predicts the density of influenced users at different levels from the source node. The model has validated on a dataset collected from Twitter. RADDIFF model can accurately predict the information diffusion within a community. The results show that the RADDIFF density reaches high number of users in the networks even though the initial set of users relatively small.