POLAR: A Benchmark for Multilingual, Multicultural, and Multi-Event Online Polarization

TL;DR

POLAR is a comprehensive multilingual dataset designed to analyze online polarization across cultures and events, revealing the complexity and contextual nature of polarization through model evaluations.

Contribution

The paper introduces POLAR, a novel multilingual, multicultural, multi-event dataset with detailed polarization annotations, and evaluates language models on this challenging benchmark.

Findings

Models perform well in binary polarization detection

Performance drops in predicting polarization types and manifestations

Highlights the contextual complexity of online polarization

Abstract

Online polarization poses a growing challenge for democratic discourse, yet most computational social science research remains monolingual, culturally narrow, or event-specific. We introduce POLAR, a multilingual, multicultural, and multi-event dataset with over 110K instances in 22 languages drawn from diverse online platforms and real-world events. Polarization is annotated along three axes, namely detection, type, and manifestation, using a variety of annotation platforms adapted to each cultural context. We conduct two main experiments: (1) fine-tuning six pretrained small language models; and (2) evaluating a range of open and closed large language models in few-shot and zero-shot settings. The results show that, while most models perform well in binary polarization detection, they achieve substantially lower performance when predicting polarization types and manifestations. These findings highlight the complex, highly contextual nature of polarization and demonstrate the need for robust, adaptable approaches in NLP and computational social science. All resources will be released to support further research and effective mitigation of digital polarization globally.