Inadequacy of SIR Model to Reproduce Key Properties of Real-world Spreading Phenomena: Experiments on a Large-scale P2P System

Daniel Bernardes, Matthieu Latapy, Fabien Tarissan

In Journal of Social Network Analysis and Mining, 3(4):1195-1208,Springer, 2013.

Understanding the spread of information on complex networks is a key issue from a theoretical and applied perspective. Despite the effort in developing theoretical models for this phenomenon, gauging them with large-scale real-world data remains an important challenge due to the scarcity of open, extensive and detailed data. In this paper, we explain how traces of peer-to-peer file sharing may be used to this goal. We reconstruct the underlying social network of peers sharing content and perform simulations on it to assess the relevance of the standard SIR model to mimic key properties of real spreading cascades. First we examine the impact of the network topology on observed properties. Then we turn to the evaluation of two heterogeneous extensions of the SIR model. Finally we improve the social network reconstruction, introducing an affinity index between peers, and simulate a SIR model which integrates this new feature. We conclude that the simple, homogeneous model is insufficient to mimic real spreading cascades. Moreover, none of the natural extensions of the model we considered, which take into account extra topological properties, yielded satisfying results in our context. This raises an alert against the careless, widespread use of this model.