Network polarization, filter bubbles, and echo chambers: An annotated review of measures and reduction methods

TL;DR

This paper reviews various measures of network polarization and discusses methods for reducing it, including modifications to network structure and recommendation systems, to address echo chambers and social fragmentation.

Contribution

It provides an annotated overview of existing polarization measures and reduction strategies, integrating approaches from multiple disciplines.

Findings

Multiple measures based on homophily, modularity, and random walks are used to quantify polarization.

Reduction strategies include edge/node edits, social network redesign, and recommendation system adjustments.

The review highlights interdisciplinary approaches to understanding and mitigating polarization.

Abstract

Polarization arises when the underlying network connecting the members of a community or society becomes characterized by highly connected groups with weak inter-group connectivity. The increasing polarization, the strengthening of echo chambers, and the isolation caused by information filters in social networks are increasingly attracting the attention of researchers from different areas of knowledge such as computer science, economics, social and political sciences. This work presents an annotated review of network polarization measures and models used to handle the polarization. Several approaches for measuring polarization in graphs and networks were identified, including those based on homophily, modularity, random walks, and balance theory. The strategies used for reducing polarization include methods that propose edge or node editions (including insertions or deletions, as well as edge weight modifications), changes in social network design, or changes in the recommendation systems embedded in these networks.