Exploiting the time-dynamics of news diffusion on the Internet through a generalized Susceptible-Infected model

We construct a news spreading model with a time dependent contact rate which generalizes the classical Susceptible–Infected model of epidemiology. In particular, we are interested on the time-dynamics of the sharing and diffusion process of news on the Internet. We focus on the counting process describing the number of connections to a given website, characterizing the cumulative density function of its inter-arrival times. Moreover, starting from the general form of the finite dimensional distribution of the process, we determine a formula for the time-variable rate of the connections and establish its relationship with the probability density function of the interarrival times. We finally show the effectiveness of our theoretical framework analyzing a real-world dataset, the Memetracker dataset, whose parameters characterizing the diffusion process are determined.