Flockworks: A Class of Dynamic Network Models for Face-to-Face Interactions

Studying the dynamics of face-to-face interaction networks is essential for a better understanding of contact mediated phenomena, e.g. contagion processes, and disease spreading [1]. Based on high-resolution proximity sensor data, recent studies analyzed the statistical properties of time-dependent contact networks [2,3], revealing heterogeneous distributions of inter-activity intervals, persistence time of groups and their size distribution, and typically strong circadian modulations. Generative network models have been proposed to capture these broad distributions [3], but require heterogeneous inter-activity times as a dynamic ingredigent and do not account for circadian node activity. We introduce a class of Markovian minimal dynamic network models that naturally yield group formation and are easy to control with a small number of natural parameters. We show that neither node heterogeneity nor postulated inter-activity time distributions are required as node or link internal properties to replicate the statistical properties of a variety of empirical systems ranging from contacts between conference attendees to high school students. We show that a number of fat-tailed statistical properties of time-dependent contact networks are a consequence of temporal modulations and show that a variety of network properties can be computed. Because of the structural simplicity of Flockworks, the model is an ideal candidate as a reference model for empirical time-dependent network data, e.g. to investigate dynamical processes that unfold on these networks.