Anomalous information diffusion in social networks: Twitter and Digg

Abstract

Diffusion is an important phenomenon in different sciences like epidemiology, economy, biology and social sciences. Information diffusion is mainly important for two reasons, to study and forecast diffusion growth and to find the source of information and validate information clarity. Researchers study diffusion in social networks and observe different aspects of them which affect diffusion properties but, diffusion type of information diffusion is not studied before. Diffusion type is normal or anomalous which have different properties and should be modeled differently, for example, random walk in one-dimensional lattice is normal diffusion but what about random walk on networks? In this paper diffusion type of discrete time random walk (DTRW) and continuous time random walk (CTRW) over 1D lattice and simple graph are studied also, diffusion type of two social networks namely Twitter and Digg are studied. Anomalous diffusion is important since most of the systems in nature have normal diffusion and there are a few numbers of systems with anomalous diffusion. To the best of our knowledge this the first time where diffusion type of information diffusion in social networks is addressed. Our results show that information diffusion in both of the Digg and Twitter is anomalous and to be more specific it is subdiffusion but, by looking more deeply, diffusion type is superdiffusion in early hours after starting diffusion in some diffusions. Some strategies are introduced to make anomalous diffusions closer to normal diffusion, which is important for policy makers of health and marketing agencies.