Hate speech detection in the Bengali language: a comprehensive survey (Journal of Big Data)

There hasn’t been much research done on techniques for identifying HS in languages other than English, like Bengali. The survey lists the main difficulties unique to Bengali HS detection, such as the dearth of labeled datasets, linguistic subtleties, and contextual differences. It looks at various methods and approaches, such as ensemble approaches, traditional machine learning techniques, and more recent developments in deep learning. In order to shed light on the efficacy of the suggested models, it also examines the performance metrics that were employed for assessment, such as accuracy, precision, recall, receiver operating characteristic (ROC) curve, area under the ROC curve (AUC), sensitivity, specificity, and F1 score. Furthermore, the limitations and future directions of research in Bengali HS detection were identified. To improve the detection accuracy, we highlighted the need for larger annotated datasets, crosslingual transfer learning techniques, and the incorporation of contextual information.

https://journalofbigdata.springeropen.com/articles/10.1186/s40537-024-00956-z