On the usage of epidemiological models for information diffusion over twitter

The way information spreads through online social networks is popularly considered to be similar to how viruses spread through a population. In this work, we study the suitability of using epidemiological models to model the spread of hashtags over Twitter by analyzing the nature of the spread. First, we define two extensions of the popular SIR model called Exo-SIR and Exo-SIS and their variants and study all the prominent hashtags in the dataset. Then, we study the hashtags that are about events and that are not about events separately. We found that the predominant nature of the spread of information over Twitter is endogenous. However, it is exogenous in the absence of events. The predominant nature of the spread of epidemics is endogenous. This implies that the usage of epidemiological models to model spread of information over Twitter is appropriate only if the spread is during events.