BD-FakeDetect: A Bangla Misinformation Dataset
Published: 10 October 2025| Version 1 | DOI: 10.17632/mc9n3gpw38.1
Contributors:
, , , , Description
BD-FakeDetect is a manually curated Bangla misinformation dataset of 5,929 news/posts from trusted fact-checkers (Rumor Scanner, FactWatch, BOOM Bangladesh, Jachai) with rich text+image metadata. Each entry provides title/content/source/date/claim-type/context/final-decision/summary, an 8-class label (False, Misleading, Altered, Scam, Satire, True, Partially True/False), and image paths for multimodal research.
Files
Steps to reproduce
1. Install: pip install pandas pyarrow 2. Load: df = pd.read_parquet("BD-FakeDetect.parquet") 3. Check: len(df)==5929 and labels ∈ {False, Misleading, Altered, Scam, Satire, True, Partially True, Partially False}.
Institutions
- University of Liberal Arts Bangladesh
Categories
Computer Vision, Natural Language Processing, Bengali Language, Multimodality