Memory in network flows and its effects on community detection, ranking, and spreading
Andrea Lancichinetti, Ice Lab, Umeň University, Umeň, Sweden
Random walks on networks is the standard tool for modelling spreading processes in social and biological systems. However, this conventional first-order Markov approach ignores a potentially important feature of the dynamics: where flow moves to may depend on where it comes from. Here we analyse multi-step pathways from different systems and show that ignoring the effects of second-order Markov dynamics has important consequences for community detection and ranking, but only marginal consequences for disease spreading through air travel. When people travelling by plane can transmit infections to random other people in cities, the memory effects from their travel patterns are lost. Therefore, accurately modelling air travel patterns has a negligible effect on disease spread. However, when focusing on the travel patterns themselves, or on systems with limited mixing, we observed that random flow on networks understates the effect of communities and exaggerates the effect of highly connected nodes. For example, capturing dynamics with a second-order Markov model allows us to differentiate airport hubs from popular destinations and reveal actual travel patterns in air traffic, and to uncover multidisciplinary journals and ranking that favour specialized journals in scientific communication. These findings were achieved only by using more available data and making no additional assumptions, and therefore suggest that accounting for higher-order memory in network flows can help us better understand how real systems are organized and function.