Real_Fake_and_Not_News is a large, balanced Bangla-language corpus designed for three-class text classification. It combines verified real news, fake news, and general Bangla sentences into a single, unified dataset—enabling classification models to learn both news authenticity and the boundary between news-like and non-news text.
The dataset is intended for researchers working on fake news detection, Bangla natural language processing (NLP), and text classification for low-resource languages.