Abstract: The present research utilises the BHSSP dataset, which has 50,000 labelled instances, to use machine learning and deep learning models to offer a hybrid architecture for Bengali hate speech ...