Figure 2

Differences in neighbourhood overlap (top and bottom) and selected temporal features of human communication (columns) for ties of similar communication intensity (\(w \in [40, 55)\)). Weak (top) and strong (bottom) ties are defined by low and high overlap, respectively, where we analyze communication patterns of red ties (ij), blue ties represent common neighbors (\(\mathcal{N}_{i} \cap \mathcal{N}_{j}\)), and grey ties represent neighbors of either i or j but not both. (Left) IET distribution; we model the time between two consecutive calls as a homogeneous process via a random variable τ, and obtain statistics on the IET distribution, such as τ̄ and \(\sigma _{\tau }\), both measured in days. (Center) Bursty Cascades; given a parameter Δt, we identify event bursts E as sub-sequences of calls that are placed at most Δt seconds within each other. \(N^{E}\) is the number of events. (Right) Temporal stability; we focus on the first and last events and their distribution within the observation period (red dashed lines), and determine the \(age\) as the time until the first call, the temporal stability TS as the time window where we observed events, and freshness f as the time between the last event and the end of the observation period