LLM-Based Multi-Task Bangla Hate Speech Detection: Type, Severity, and Target

ACL ARR 2026 January Submission7147 Authors

Keywords: Bangla Hate speech detection, datasets for low resource languages, benchmarking, hate speech detection, model evaluation

Abstract: Online social media platforms have become central to communication and information exchange, however, they also serve as fertile ground for hate speech, offensive language, and bullying targeting individuals and communities. Such content undermines online safety and inclusion, underscoring the need for reliable detection systems—especially in low-resource languages with limited moderation tools. For Bangla, existing work provides valuable resources and models, however, they are mostly single-task (e.g., binary hate/offense) with narrow coverage of key dimensions such as type, severity, and target. We address these gaps by introducing \textit{the first multi-task} Bangla hate-speech dataset, BanglaMultiHate, one of the largest manually annotated dataset to date. Using this resource, we performed a comparative study across different baselines, monolingual pretrained models, and LLMs under zero-shot and LoRA fine-tuning settings. Our findings show that while LoRA-tuned LLMs rival BanglaBERT, culturally grounded pretraining remains crucial for robust performance. Overall, BanglaMultiHate establishes a stronger benchmark for hate speech detection in low-resource contexts. All data and scripts will be released for reproducibility.

Paper Type: Long

Research Area: Resources and Evaluation

Research Area Keywords: datasets for low resource languages, benchmarking, hate speech detection, model evaluation

Contribution Types: Model analysis & interpretability, Approaches to low-resource settings, Publicly available software and/or pre-trained models, Data resources, Data analysis

Languages Studied: Bangla

Submission Number: 7147