Hate Speech and Offensive Language Detection in Bengali

TL;DR

This paper introduces a new Bengali hate speech dataset, explores baseline models and transfer learning techniques for classification, and finds that models like XLM-Roberta and MuRIL excel in detecting offensive content in both native and Romanized Bengali.

Contribution

It creates the first large-scale annotated Bengali hate speech dataset including Romanized text and evaluates multiple models with transfer learning for improved detection.

Findings

XLM-Roberta performs best on separate datasets.

MuRIL outperforms others in joint and few-shot training.

Code and dataset are publicly available.

Abstract

Social media often serves as a breeding ground for various hateful and offensive content. Identifying such content on social media is crucial due to its impact on the race, gender, or religion in an unprejudiced society. However, while there is extensive research in hate speech detection in English, there is a gap in hateful content detection in low-resource languages like Bengali. Besides, a current trend on social media is the use of Romanized Bengali for regular interactions. To overcome the existing research's limitations, in this study, we develop an annotated dataset of 10K Bengali posts consisting of 5K actual and 5K Romanized Bengali tweets. We implement several baseline models for the classification of such hateful posts. We further explore the interlingual transfer mechanism to boost classification performance. Finally, we perform an in-depth error analysis by looking into the misclassified posts by the models. While training actual and Romanized datasets separately, we observe that XLM-Roberta performs the best. Further, we witness that on joint training and few-shot training, MuRIL outperforms other models by interpreting the semantic expressions better. We make our code and dataset public for others.