Fake News Dataset for Text Classification

Fake news articles and labels for detecting misinformation, curated by author noahgift. The dataset contains between 1,000 and 10,000 text records and was last updated in June 2022.

Use Cases

• Classify text articles as 'fake' or 'real' using labeled categories.
• Train a BERT model to predict article veracity from textual content and metadata.
• Analyze linguistic patterns and bias in misinformation articles.
• Build a binary classifier for news authenticity using text features.

Strengths

• Dataset size is confirmed to be between 1,000 and 10,000 rows.
• Text modality is explicitly suited for NLP tasks.

Limitations

• Specific column names and data structure are unknown, complicating direct analysis.
• Temporal coverage and geographic scope of the articles are unspecified.
• Potential class imbalance or label noise cannot be assessed without sample data.

Provenance

Source

Hugging Face, author noahgift.

Collection Method

Curated collection of news articles with labels; exact gathering method unknown.

Time Range

null

Freshness

Last updated in 2022; may not include recent misinformation trends.

Geography

Region tag indicates 'us', suggesting a US focus, but not confirmed.