EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

BanglaHateCorpus: A large-scale bangla benchmark for hate speech classification and a transformer-based deep learning method

Anika Tasnim Rodela iD, Mahbub E Sobhani, Sakif Naieb Raiyan iD, Mohammad Nurul Huda, Dewan Md Farid iD

PLOS One · 04.11.2026 · 21 (11) · e0361339

DOI: 10.1371/journal.pone.0361339 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 04.11.2026. Verfasst von Anika Tasnim Rodela und 4 weiteren Autor:innen. Beteiligte Einrichtung: Department of Information and Communication Technology, Bangladesh University of Professionals, Dhaka, Bangladesh. Hauptthemen laut MEDLINE: Deep Learning, Hate, Social Media, Speech. Die Arbeit nennt 77 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Information and Communication Technology Division.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Hate speech on social media refers to content that threatens, attacks, or dehumanizes people or groups based on specified characteristics, thereby eroding trust, worsening societal divisions, and causing psychological harm to the targeted communities. In digitally connected South Asian societies, strong in-group bonds and online inter-group hatred can rapidly lead to real-world violence, a risk intensified by the proliferation of Bangla texts, highlighting the urgent need for robust detection in this low-resource language. Previous studies have mostly focused on building their datasets by scraping various sources, without emphasizing standardized quality annotation or ensuring classification accuracy. To address these gaps, we introduce BanglaHateCorpus, a unified corpus of 168K examples consolidated from eight public sources, rigorously de-duplicated and annotated through a two-stage, human and LLM-assisted pipeline. Building on this benchmark, we propose BanHateFormer, a novel deep-learning model that embodies state-of-the-art transformer components, and replaces the conventional feed-forward blocks with a Mixture-of-Experts layer followed by a binary classification head. BanHateFormer achieves state-of-the-art classification results, as assessed by accuracy, precision, recall, and macro F1-score with approximately 94.74 times fewer parameters than comparable models, achieving a 91.33% macro F1-score on the BanglaHateCorpus dataset and 78.96% on the transliterated BanTH benchmark. BanHateFormer outperforms BanglaBERT and XLM-RoBERTa by up to 2.57% with three times fewer parameters, and improves macro F1-score by 3.38% on transliterated text, demonstrating strong performance on romanized inputs. BanHateFormer significantly surpasses existing transformer-based models in balancing memory efficiency and high-fidelity outputs, marking a breakthrough in hate speech classification with optimal space utilization.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningHateSocial MediaSpeechHumansLarge Language Models

Autor:innen und Institutionen

  • Department of Information and Communication Technology, Bangladesh University of Professionals, Dhaka, Bangladesh.
  • Department of Computer Science & Engineering, United International University, Dhaka, Bangladesh.
  • Department of Computer Science and Engineering, BRAC University, Dhaka, Bangladesh.
  • Department of Computer Science and Engineering, Bangladesh University of Engineering & Technology, Dhaka, Bangladesh.
  • Department of Computer Science & Engineering, Southeast University, Dhaka, Bangladesh.

ORCID-Profile: Anika Tasnim Rodela, Sakif Naieb Raiyan, Dewan Md Farid

Förderung

  • Information and Communication Technology Division

Zitieren

APA

Rodela, A. T., Sobhani, M. E., Raiyan, S. N., Huda, M. N., & Farid, D. M. (2026). BanglaHateCorpus: A large-scale bangla benchmark for hate speech classification and a transformer-based deep learning method. PLOS One, 21(11), e0361339. https://doi.org/10.1371/journal.pone.0361339

Vancouver

Rodela AT, Sobhani ME, Raiyan SN, Huda MN, Farid DM. BanglaHateCorpus: A large-scale bangla benchmark for hate speech classification and a transformer-based deep learning method. PLOS One. 2026;21(11):e0361339. doi:10.1371/journal.pone.0361339

BibTeX
@article{Rodela2026,
  author = {Rodela, Anika Tasnim and Sobhani, Mahbub E and Raiyan, Sakif Naieb and Huda, Mohammad Nurul and Farid, Dewan Md},
  title = {BanglaHateCorpus: A large-scale bangla benchmark for hate speech classification and a transformer-based deep learning method},
  journal = {PLOS One},
  year = {2026},
  volume = {21},
  number = {11},
  pages = {e0361339},
  doi = {10.1371/journal.pone.0361339}
}
Kostenlos starten →