A Multi-Labeled Dataset for Indonesian Discourse: Examining Toxicity, Polarization, and Demographics Information

Online discourse is increasingly trapped in a vicious cycle where polarizing language fuelstoxicity and vice versa. Identity, one of the most divisive issues in modern politics, oftenincreases polarization. Yet, prior NLP research has mostly treated toxicity and polarization asseparate problems. In Indonesia, the world’s third-largest democracy, this dynamic threatens democratic discourse, particularly in online spaces. We argue that polarization and toxicity must be studied in relation to each other. To this end, we present a novel multi-label Indonesian dataset annotated for toxicity, polarization, and annotator demographic information. Benchmarking with BERT-base models and large language models (LLMs) reveals that polarization cues improve toxicity classification and vice versa. Including demographic context further enhances polarization classification performance.

Citation

L. Susanto, M.I. Wijanarko, P.A. Pratama, Z. Tang, F. Akyas, T. Hong, I.K. Idris, A.F. Aji, D.T. Wijaya, "A Multi-Labeled Dataset for Indonesian Discourse: Examining Toxicity, Polarization, and Demographics Information," 2025, pp. 18863-18890.

Conference

Findings of the Association for Computational Linguistics: ACL 2025

Keywords

44 Human Society, 4408 Political Science, 47 Language, Communication and Culture

Source

Findings of the Association for Computational Linguistics: ACL 2025

Publisher

Association for Computational Linguistics

DOI

10.18653/v1/2025.findings-acl.966

Additional links

https://aclanthology.org/2025.findings-acl.966/