Fig. 1: Construction of temporal networks from intermittent social interactions.

a Social interactions between \(8\) individuals indicated by solid circles with different colours. Along the whole time from \(t = 1\) to \(t = {\Bbb T}\), each individual is depicted by the same colour line, over which the corresponding circles will be given and connected with each other at time \(t\) provided two players interact with each other during the time interval \((t - \tau ,t]\). Here \(\tau = 1\) for the simplicity of visualisations, and normally in the real data collected by SocioPatterns (see Methods), \(\tau = 20\)s. b Four different temporal networks that arise from aggregating the interactions shown in a into snapshots using different time windows \(\Delta t\). When \(\Delta t = {\Bbb T}\), all interactions are captured in a single snapshot, corresponding to the static network that is the typical object of study in social network data. In general, when \(\Delta t \, < \, {\Bbb T}\), we have \(\left\lceil {{\Bbb T}/\Delta t} \right\rceil\) snapshots. c The definition of evolutionary process on temporal networks. Taking the temporal network corresponding to \(\Delta t = 4\) in (b) as an example, we perform \(g\) rounds of evolution in each snapshot before changing the network structure to the next one, and totally we run \(G\) rounds. If \(\left\lceil {{\Bbb T}/\Delta t} \right\rceil g \, < \, G\), we repeat the sequence of snapshots from the beginning.