Khasi cyberbullying detection datasets

A manually collected and annotated Khasi-language social media dataset for cyberbullying detection. The comments were gathered from public pages and posts and annotated by two native Khasi speakers to ensure accurate interpretation of slang, context, and cultural nuances. The dataset reflects real-world online communication and is suitable for NLP research, classification tasks, and low-resource language processing.
Labels

Bullying: Comments containing abusive, offensive, insulting, or threatening language.(3,268 comments)
Non-bullying: Neutral, friendly, or informative comments without any harmful intent.(5,342 comments)

Total Records

8,610 comments (combined from Facebook, Instagram, and YouTube)