Title of Flavio’s talk: “Effective distances in complex networks and influencer identification“

The large amount of data sets that became available in recent years has made it possible to empirically study humanly-driven, as well as biological complex systems to an unprecedented extent. In parallel, the prediction and control of epidemic outbreaks have become very important for public health issues. In this talk, we discuss some important aspects of diffusion phenomena and spreading processes unfolding on complex networks. The first problem is the estimation of the arrival times of epidemics on the global scale using the metapopulation-network approach. To this end, we derive and identify suitable hidden geometries, leveraging on random-walk theory. Through the definition of network ‘effective distances’, the problem of complex spatiotemporal patterns is reduced to simple, homogeneous wave propagation patterns. Secondly, by embedding nodes in the hidden space defined by network effective distances, we introduce a novel network centrality, called ‘ViralRank’, which quantifies how close a node is, on average, to the other nodes. As a case study, we first characterize the political leanings and, using known heuristic centralities, rumor spreading dynamics on two networks built on datasets extracted from Twitter. Then, we investigate the role of centrality measures in identifying influential spreaders by comparing the relative performance with ViralRank in several empirical datasets of social, biological and infrastructure complex systems. We find that ViralRank can correctly identify influential nodes in the supercritical regime for both contact networks and metapopulations, as it systematically outperforms state-of-the-art centrality measures.