Hate speech detection in social media

Abstract

Hate speech has been defined as the act of offending, insulting, or threatening individuals or a group of people based on their religion, race, caste, orientation, gender, or belongingness to a specific stereotyped community leading to paranoia in society. The exponential rise in the use of online social media has led to an increase in the use of hate speech online. The use of code-mixed languages on these media channels has made the problem of detecting hate speech even more arduous, especially in multilingual societies like India. Adding to all these issues, most social media applications are conversational based. This leads to the problem of the absence of context from text unless added. In this paper, we propose a transformer-based model to detect hate speech in conversational code-mixed data. The model proposed outperforms most of the models with minimal preprocessing required on the datasets.