Towards the development of the Bengali language corpus from public Facebook pages for hate speech research

Abstract

Online abusive or hateful speech detection in different languages on Social Networking Sites (SNS) have drawn the attention of researchers recently. Hateful comments in public Facebook pages ignite social mishaps in Bangladesh. In this paper, we have discussed the development and annotation of the corpus of hateful speech in the Bengali language on public Facebook pages. We have classified hateful comments into six major classes based on the social aspects of Bangladesh. The corpus (4753 comments) is the maiden contribution as a publicly available data set that can be enhanced and utilized for future hate speech research in SNS.