Dresden 2020 – wissenschaftliches Programm

Die DPG-Frühjahrstagung in Dresden musste abgesagt werden! Lesen Sie mehr ...

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

SOE: Fachverband Physik sozio-ökonomischer Systeme

SOE 11: Networks - From Topology to Dynamics I (joint SOE/DY/BP)

SOE 11.2: Vortrag

Mittwoch, 18. März 2020, 17:45–18:00, GÖR 226

Inferring political spaces from retweet networks — •Eckehard Olbrich, Felix Gaisbauer, Armin Pournaki, and Sven Banisch — Max Planck Institute for Mathematics in the Sciences, Leipzig, Germany, Leipzig,

When people analyze retweet networks from Twitter it is quite common to interpret visualizations generated by some force-directed layout algorithm as political spaces in the sense that distances between nodes are interpreted as political distances. We investigate, under which conditions this is a valid interpretation by building a statistical model for retweet networks in political spaces, i.e. where the agents have a position in a metric space and the retweet probability depends on their distance. These models can be considered as extensions of spatial random graph models for social networks, for which inference algorithms of the underlying social spaces are known [1]. We show that force-directed layout algorithms can be related to maximum likelihood estimators of theses models by using gradient decent. Finally, we propose layout algorithms that are specifically adapted to the task of embedding the graph in a political space.

[1] P. D. Hoff, A. E. Raftery, and M. S. Handcock (2002). Latent space approaches to social network analysis. Journal of the american Statistical association, 97(460), 1090-1098.