AG News Corpus for Text Classification

AG News is a text classification dataset containing over 100,000 news articles. SetFit created this dataset, which was last updated on Hugging Face in January 2022. It is categorized for news aggregation and classification tasks.

Use Cases

• Classify article topics using the provided text and labels.
• Train models for news headline categorization based on article content.
• Perform text analysis on news article features like title and description.
• Benchmark multi-class text classifiers on a labeled news corpus.

Strengths

• Contains over 100,000 news articles, as indicated by platform size tags.
• Provides labeled data suitable for supervised text classification tasks.

Limitations

• Dataset has not been updated since January 2022, limiting temporal relevance.
• Specific column names and data schema are not documented in the provided input.
• Potential geographic or source bias is unknown due to lack of provenance details.

Provenance

Source

SetFit

Collection Method

News aggregation and labeling, specific method unknown.

Time Range

null

Freshness

Last updated 2022-01-19.

Geography

Region tags suggest US focus, but exact coverage is unknown.