Hyperpartisan News Articles for Media Bias Analysis

Hyperpartisan News is a text dataset for media bias and political stance classification. The dataset was created by user 'zapsdcn' and last updated on Hugging Face in December 2021. It is categorized as containing 1K to 10K samples and is tagged for US regional focus.

Use Cases

• Classify article bias (hyperpartisan vs. non-hyperpartisan) using the text content and associated labels.
• Analyze linguistic patterns and framing in political news articles using the full text field.
• Train models to predict the political leaning of media outlets based on their published article text.
• Study the correlation between article topics and hyperpartisan labeling within the dataset.

Strengths

• Dataset size is categorized between 1,000 and 10,000 text samples.
• Focus on a specific, well-defined task: hyperpartisan news detection.

Limitations

• Sample size is relatively small for training complex deep learning models.
• Geographic scope appears limited to US media, introducing regional bias.
• Data is stale, last updated over two years ago, which may not reflect current media landscape.

Provenance

Source

Hugging Face dataset uploaded by user 'zapsdcn'.

Collection Method

Collection method unknown; likely involves scraping or curating news articles.

Time Range

null

Freshness

Last updated on 2021-12-08; no update frequency stated.

Geography

Tagged for US region ('Regionus').