30,000 to 50,000 labeled English news articles categorized for text classification and fact-checking. The dataset consists of original monolingual content designed for supervised learning in the domain of misinformation detection.
Use Cases
- Train a binary classifier to identify misinformation using the text and label fields
- Evaluate fact-checking models on a corpus of over 30,000 news articles
- Perform intent classification to distinguish reporting styles using the text column
Strengths
- Contains a volume of 30,000 to 50,000 text samples
- Features task-specific labels for fact-checking and intent-classification
- Sourced as original data for monolingual English (en) analysis