Preferential attachment in face-to-face relations - Archive ouverte HAL Accéder directement au contenu
Pré-Publication, Document De Travail Année : 2024

Preferential attachment in face-to-face relations

Résumé

We propose a stochastic model to describe the dynamics of contact duration in face to face interactions. It reproduces precisely the deviations from the mean-time that were found to be statistically very similar in several environments. The model is based on geometric graphs built from points of a Levy walk. It has a single parameter, the Levy index, that is found to be $\alpha\simeq1.1$ on all the datasets. In this model, the duration of contacts in face to face interactions, which corresponds to the length of discussions for humans, is related to the size of the connected components (clusters) of the Levy graph. We compute analytically the distribution of the clusters size which is of stretched-exponential type and reveals the dynamics of the process. A discussion starts about a topic that is first developed, which, in the Levy framework is described by a geometric distribution. It is then driven by a preferential attachment mechanism; the more a discussion lasts, the more it is likely to continue since there are more points to come back to. This leads to heavy tails in the contact duration. The combined behavior is well captured by the formation of clusters in Levy graphs. We show it has some similarities with the case of a growing network with a sub-linear power-law kernel. More precisely, in this model discussions appear as an aggregation process with a natural scale fixed by the mean interaction time between the individuals.

Dates et versions

hal-04523743 , version 1 (27-03-2024)

Identifiants

Citer

S. Plaszczynski, B. Grammaticos, M. Badoual. Preferential attachment in face-to-face relations. 2024. ⟨hal-04523743⟩
0 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More