Elamur Research⌕
☰
home / papers / 2026 / From Comments to Harm: A Findings Report on Abusive Tamil Text Targeting Women on Social Media Shared Task

From Comments to Harm: A Findings Report on Abusive Tamil Text Targeting Women on Social Media Shared Task

Paper · 2026 · Bhuvaneswari Sivagnanam, Kathiravan Pannerselvam, Jananayagan et al.

Summary

This report covers the second edition of a shared task on detecting Tamil social-media comments that abuse women. It was held at DravidianLangTech@ACL 2026 and hosted on Codabench. Online harassment often relies on sarcasm, stereotypes and indirect wording. Tamil adds further difficulty: it is low-resource, spelling is informal and varies by region, and writers often mix Tamil with English or use Romanised script. The organisers built a corpus of Tamil YouTube comments from entertainment, political and social videos, combining automated scraping with manual filtering. Each comment carries a binary label. A comment counts as Abusive only when a woman is the target, so insults aimed at men stay Non-Abusive even if they mention female relatives. Systems were ranked by macro-F1 to reward balanced performance on both classes. The top entry reached 0.8297 macro-F1. Fine-tuned encoders matched to the language and domain performed reliably well, and Tamil-specific normalisation and calibration aimed at macro-F1 brought further gains. The benchmark offers a shared basis for comparing gender-focused moderation methods in Dravidian languages.

Key points

  • Participation: 89 teams registered and 24 submitted results. Most scores fell roughly between 0.79 and 0.81 macro-F1, with the best at 0.8297.
  • Annotation: three native Tamil speakers labelled the data. Two studied Computer Science and one Social Work; two were Ph.D. students and one a postgraduate. The mix of backgrounds was meant to reduce subjective bias.
  • Methods submitted: transformer fine-tuning, heterogeneous ensembles, classical baselines, and LLMs used through prompting or LoRA adaptation.
  • What helped most: class-weighted training, threshold tuning on validation data, and preprocessing that handles obfuscated or masked abuse.
  • Bottom of the leaderboard: several teams tied at rank 20. Their outputs did not all collapse to one class; the shared score came from different problems, such as majority-class bias or skewed decision thresholds.
  • Limitations: the data comes only from YouTube, and the binary labels may miss sarcasm, implicit abuse and counterspeech. The authors suggest finer-grained annotation for future editions.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

 Institutions

🎓Ollscoil na Gaillimhe – University of GalwayGalway · Ireland23 works · 2,229 citations🎓Vellore Institute of Technology UniversityVellore · India7 works · 268 citations🎓Central University of Tamil NaduThiruvarur · India3 works · 15 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 works

Cite

Cite as Bhuvaneswari Sivagnanam, Kathiravan Pannerselvam, Jananayagan, Charmathi Rajkumar, Ramesh Kannan R, Ratnavel Rajalakshmi, Shunmuga Priya Muthusamy Chinnan, Saranya Rajiakodi and Bharathi Raja Chakravarthi. 2026. From Comments to Harm: A Findings Report on Abusive Tamil Text Targeting Women on Social Media Shared Task. Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2026.dravidianlangtech-1.11

@inproceedings{sivagnanam2026comments,
  title = {{From Comments to Harm: A Findings Report on Abusive Tamil Text Targeting Women on Social Media Shared Task}},
  author = {Bhuvaneswari Sivagnanam, Kathiravan Pannerselvam, Jananayagan, Charmathi Rajkumar, Ramesh Kannan R, Ratnavel Rajalakshmi, Shunmuga Priya Muthusamy Chinnan, Saranya Rajiakodi and Bharathi Raja Chakravarthi},
  booktitle = {Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages},
  year = {2026},
  doi = {10.18653/v1/2026.dravidianlangtech-1.11},
  url = {https://doi.org/10.18653/v1/2026.dravidianlangtech-1.11}
}

References

40 references, 9 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Rameez Mohammed A and S D Madhu Kumar. 2026. NITC-HSR@DravidianLangTech 2026: Ensembling Multilingual Transformer Models for Detecting Abusive Tamil Text Targeting Women on Social Media. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. Associ… find
  2. Fitsum Gaim, Hoyun Song, Huije Lee, Changgeon Ko, Eui Jun Hwang, and Jong C. Park. 2025. A multi-task benchmark for abusive language detection in low-resource settings. Preprint, arXiv:2505.12116.
  3. Vaishali Ganganwar and Ratnavel Rajalakshmi. 2022. Mtdot: A multilingual translation-based data augmentation technique for offensive content identification in tamil text data. Electronics, 11(21). find
  4. Valerio Basile, Cristina Bosco, Elisabetta Fersini, Debora Nozza, Viviana Patti, Francisco Rangel Pardo, Paolo Rosso, and Manuela Sanguinetti. 2019. Semeval-2019 task 5: Multilingual detection of hate speech against immigrants and women in twitter. In Proceedings of the 13th International Workshop o…
  5. Yuting Guo, Sangmi Kim, Elise Warren, Yuan-Chi Yang, Sahithi Lakamana, and Abeed Sarker. 2023. Automatic detection of intimate partner violence victims from social media for proactive delivery of support. find
  6. In AMIA Joint Summits on Translational Science Proceedings, pages 254–260. American Medical Informatics Association. find
  7. Giacomo Marzi, Marco Balzano, and Davide Marchiori. 2024. K-alpha calculator–krippendorff’s alpha calculator: a user-friendly tool for computing krippendorff’s alpha inter-rater reliability coefficient. MethodsX, 12:102545. find
  8. Maria Cristina Hinojosa Lee, Johan Braet, and Johan Springael. 2024. Performance metrics for multilabel emotion classification: comparing micro, macro, and weighted f1-scores. Applied Sciences, 14(21):9863. find
  9. Sandra Lopes Miranda. 2023. Analyzing hate speech against women on instagram. Open Information Science, 7(1):20220161. find
  10. Md. Refaj Hossan, Kawsar Ahmed, and Mohammed Moshiul Hoque. 2025. CUET-NLP_Zenith at BLP-2025 task 1: A multi-task ensemble approach for detecting hate speech in Bengali YouTube comments. In Proceedings of the Second Workshop on Bangla Language Processing (BLP-2025), pages 443– 452, Mumbai, India. A… find
  11. Hamdy Mubarak, Hend Al-Khalifa, and Abdulmohsen Al-Thubaity. 2022. Overview of osact5 shared task on arabic offensive language and hate speech detection. In OSACT.
  12. Nicolás Benjamín Ocampo, Ekaterina Sviridova, Elena Cabrio, and Serena Villata. 2023. An in-depth analysis of implicit and subtle hate speech messages. In Proceedings of the 17th Conference of the European Chapter of the Association for Computational Linguistics, pages 1997–2013. find
  13. Julia Jaremko, Dagmar Gromann, and Michael Wiegand. 2025. Revisiting implicitly abusive language detection: Evaluating llms in zero-shot and few-shot settings. In Proceedings of the 31st International Conference on Computational Linguistics, pages 3879– 3898. find
  14. Manish Pandey, Nageshwar Prasad Yadav, Mokshada Adduru, and Sawan Rai. 2025. Creating and evaluating code-mixed nepali-english and telugu-english datasets for abusive language detection using traditional and deep learning models. Preprint, arXiv:2504.21026. — [arXiv:2504.21026](https://arxiv.org/abs/2504.21026) find
  15. Kiruthika K, Roahiyaa T, and Premjith B. 2026. SU-PERNOVA@DravidianLangTech 2026: Transformer and Ensemble Approaches for Abusive Tamil Text Detection Targeting Women. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. Association for Computati… find
  16. Kathiravan Pannerselvam and Saranya Rajiakodi. 2026. Systematic literature review on hate speech detection in indian low-resource languages. Journal of Computational Social Science, 9(1):5. find
  17. Diya Prakash, Praveen Kumar S, Ranjith Kumar R, Sir-anjeevi Rajamanickam, Balasubramanian Palani, and Jobin Jose. 2026. DPR@DravidianLangTech 2026: Transformer-Based Abusive Content Detection for Tamil Text Targeting Women on Social Media. In Proceedings of the Sixth Workshop on Speech, Vision, and…
  18. Varalakshmi K and Bharathi B. 2026. TAMILGO-ODBADTXT@DravidianLangTech 2026:A Multilingual Transformer-Based Approach for Abusive Language Identification in Tamil Social Media. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. Association for… find
  19. Arunaggiri Pandian Karunanidhi and Prabalakshmi Arumugam. 2026. CHMOD_777@DravidianLangTech 2026: Context-Aware Fine-tuned MuRIL for Abusive Tamil Text Detection on Social Media. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. Association fo… find
  20. Kabilan Prasanna and Aryaman Arora. 2024. Irumozhi: Automatically classifying diglossia in tamil. In Findings of the Association for Computational Linguistics: NAACL 2024, pages 3096–3103. find
  21. Ruba Priyadharshini, Bharathi Raja Chakravarthi, M Subramanian, and 1 others. 2023. Overview of shared-task on abusive comment detection in tamil and telugu. In DravidianLangTech. find
  22. Emma Kavanagh, Chelsea Litchfield, and Jaquelyn Osborne. 2019. Sporting women and social media: Sexualization, misogyny, and gender-based violence in online spaces. International Journal of Sport Communication, 12(4):552–572. find
  23. Hanish Vigneshwar R, Nahul Alaguraj, Karthikeyan M, and Ratnavel Rajalakshmi. 2026. HNK@DravidianLangTech 2026: Investigating Grapheme-Level Normalization for Abusive Tamil Text Classification. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages.…
  24. Triambiga Krubhakaran, Senthil Kumar B, Kaviya Nagarajan, and Balaji N. 2026. VITECH@DravidianLangTech2026: Prompting and LoRA Adaptation for Tamil Abusive Language Detection - A Comparative Study of Open LLMs. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dra… find
  25. Saranya Rajiakodi, Bharathi Raja Chakravarthi, Shunmuga Priya Muthusamy Chinnan, Ruba Priyadharshini, Kathiravan Pannerselvam, Rahul Ponnusamy, Bhuvaneswari Sivagnanam, Paul Buitelaar, Jananayagan Jananayagan, Kishore Kumar Ponnusamy, and 1 others. 2025. Findings of the shared task on abusive tamil… find
  26. Nina Seemann, Yeong Su Lee, Julian Höllig, and Michaela Geierhos. 2023. The problem of varying annotations to identify abusive language in social media content. Natural Language Engineering, 29(6):1561– 1585. find
  27. Jacob Ratkiewicz, Michael Conover, Mark Meiss, Bruno Gonçalves, Alessandro Flammini, and Filippo Menczer. 2011. Detecting and tracking political abuse in social media. In Proceedings of the International AAAI Conference on Web and social media, volume 5, pages 297–304. find
  28. Deepawali Sharma, Vivek Kumar Singh, and Vedika Gupta. 2024. Tab hate: a target-based hate speech detection dataset in hindi. Social Network Analysis and Mining, 14(1):190. find
  29. Victoria L Stephenson, Brittany M Wickham, and Nicole M Capezza. 2018. Psychological abuse in the context of social media. Violence and Gender, 5(3):129–134. find
  30. Oarisa Rebayet, Tahmima Hoque Eid, Fawzia Tabassum, and Hasan Murad. 2026. TriVector@DravidianLangTech 2026: Abusive Tamil Text Detection on Social Media Using Lexicon-Augmented Transformers. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. A…
  31. Malliga Subramanian, Kogilavani Shanmugavadivel, Samyuktha K S, Santhiya D, Saranya P, Samritha A R, and Sivapriyan P. 2026. VI-SION@DravidianLangTech 2026: Abusive Tamil Text Detection Targeting Women on Social Media. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies… find
  32. Miftahul Jannat Rishta, Sumaiya Zaman, Shiti Chowdhury, and Hasan Murad. 2026. CUET_SYNTHETICA@DravidianLangTech 2026: Multi Architecture Transformer Ensemble for Detecting Abusive Tamil Text Targeting Women. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravi…
  33. Nicolas Suzor, Molly Dragiewicz, Bridget Harris, Rosalie Gillett, Jean Burgess, and Tess Van Geelen. 2019. Human rights by design: The responsibilities of social media platforms to address gender-based violence online. Policy & internet, 11(1):84–103. find
  34. Mirudhula Sankar and Ratnavel Rajalakshmi. 2026. DLRG@DravidianLangTech 2026: Explainable Transformer-Based Detection of Abusive Tamil Text Targeting Women on Social Media. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. Association for Comp…
  35. Rithikaa V, Sanjay Krishnan K, Nithya Varshini C N, and R S. Sumathi. 2026. Prime-Line@DravidianLangTech 2026: Abusive Tamil Comment Detection Using MuRIL. In Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages. Association for Computational Linguis… find
  36. Shrey Satapara, Prasenjit Majumder, Thomas Mandl, Sandip Modha, Hiren Madhu, Tharindu Ranasinghe, Marcos Zampieri, Kai North, and Damith Premasiri. 2023. Overview of the hasoc subtrack at fire 2022: Hate speech and offensive content identification in english and indo-aryan languages. In Proceedings… find
  37. Bertie Vidgen and Leon Derczynski. 2020. Directions in abusive language training data, a systematic review: Garbage in, garbage out. Plos one, 15(12):e0243300. find
  38. Shrey Satapara, Sarah Masud, Hiren Madhu, Md. Aflah Khan, Md. Shad Akhtar, Tanmoy Chakraborty, Sandip Modha, and Thomas Mandl. 2024. Overview of the hasoc subtracks at fire 2023: Detection of hate spans and conversational hate-speech. In Proceedings of the 15th Annual Meeting of the Forum for Inform… find
  39. Mohammad Zia Ur Rehman, Somya Mehta, Kuldeep Singh, Kunal Kaushik, and Nagendra Kumar. 2023. User-aware multilingual abusive content detection in social media. Information Processing & Management, 60(5):103450. find
  40. Bertie Vidgen, Alex Harris, Dong Nguyen, Rebekah Tromble, Scott A Hale, and Helen Margetts. 2019. Challenges and frontiers in abusive content detection. In Proceedings of the third workshop on abusive language online, pages 80–93.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/sivagnanam-2026-comments-harm-findings-report-abusive/