23,000 professionally fact-checked political statements are labeled with a 6-class credibility scale. The dataset is designed for multiclass text classification tasks, supporting models like BERT. Its origin and last update date are unknown.
Use Cases
- Classify statement_credibility into one of six labels using text features.
- Train BERT models for multiclass text classification on political discourse.
- Benchmark fake news detection algorithms against a labeled corpus of 23,000 statements.
Strengths
- 23,000 labeled text instances provide a substantial corpus for training.
- Professional fact-checking provides a reliable ground truth for the 6-class labels.
Limitations
- Unknown row count beyond the stated 23,000 statements limits precise scale assessment.
- Potential geographic or political bias is unverified without details on statement sources.
Provenance
- Collection Method
- Professional fact-checking