Berlin 2024 – scientific programme
Parts | Days | Selection | Search | Updates | Downloads | Help
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 21: Computational Social Science
SOE 21.4: Talk
Thursday, March 21, 2024, 17:45–18:00, MA 001
Hierarchical Clustering and Polarization in affiliation networks — •Emanuele Cozzo1, Adrian Fernandez Cid2, Oriol Pujol2, and Luce Prignano1 — 1Univeristat de Barcelona Institute of Complex Systems — 2Facultat de Matematiques i Informatica, Universitat de Barcelona
Social and political polarization has long been a key subject in both academic and mainstream discussions. The key focus in polarization research is its measurement. Structural methodologies emphasize deducing polarization from the characteristics of the system’s network representation. Polarization metrics necessitate explicit definitions in terms of groups, which can be identified either endogenously or exogenously. In cases involving endogenously identified groups, the challenge of measuring polarization intertwines with the task of data clustering. Commonly, data is first clustered, followed by the measurement of polarization. Yet, a more coherent approach would involve identifying clusters based on the same principles used for measuring polarization.
In our study, we investigate an axiomatic suite of polarization metrics to evaluate the polarization within the identified segments and to assist in the hierarchical clustering of social entities within the system. Our focus is particularly on affiliation networks. We demonstrate our findings using the Southern Women dataset. With the insights gleaned from our analysis of this established system, we then extend our methodology to explore contemporary, real-world situations.
Keywords: polarization; clustering; affiliation networks