This dataset was gathered from the Google Fact Checker API, using an automatic web scraper. 10,000 facts were pulled, but for the sake of simplicity, only ones were the ratings were singular words "false" or "true", were kept, which filtered it down to ~3000 fact checks, with about 90% of the facts being false.
annotations_creators:
- expert-generated language_creators:
- crowdsourced languages:
- en-US licenses:
- unknown multilinguality:
- monolingual pretty_name: polifact-covid-fact-checker size_categories:
- unknown source_datasets:
- original task_categories:
- text-classification
- question-answering task_ids:
- fact-checking
- multi-label-classification
- sentiment-classification
- closed-domain-qa
- extractive-qa