Communication dynamics in finite capacity social networks

TL;DR

This paper presents a generic model of communication in large social networks that captures the interplay between network structure and information flow, emphasizing the impact of individual temporal limitations on dynamics.

Contribution

It introduces a non-Markovian communication model based on local interactions, validated with human data, and explores the effects of finite sampling on network analysis.

Findings

Flow is non-Markovian and influenced by individual temporal constraints.

Model accurately predicts node activity and degree distribution.

Finite sampling limits affect network analysis accuracy.

Abstract

In communication networks structure and dynamics are tightly coupled. The structure controls the flow of information and is itself shaped by the dynamical process of information exchanged between nodes. In order to reconcile structure and dynamics, a generic model, based on the local interaction between nodes, is considered for the communication in large social networks. In agreement with data from a large human organization, we show that the flow is non-Markovian and controlled by the temporal limitations of individuals. We confirm the versatility of our model by predicting simultaneously the degree-dependent node activity, the balance between information input and output of nodes and the degree distribution. Finally, we quantify the limitations to network analysis when it is based on data sampled over a finite period of time.