News Headline and Body Text Stance Classification Dataset

Aggregating instances of headlines and body texts for classifying the stance of the body relative to the headline into four categories: unrelated, discuss, agree, and disagree. It was created by author nid989 and last updated in December 2021.

Use Cases

• Train a text classifier to predict the 'stance' label (agree, disagree, discuss, unrelated) from pairs of 'headline' and 'body' text.
• Analyze linguistic patterns in the 'body' text that correlate with 'agree' or 'disagree' stances towards a 'headline' claim.
• Build a model to detect if a 'body' text is 'unrelated' to a given 'headline' to filter irrelevant news article pairs.

Strengths

• Provides a defined four-category classification task (unrelated, discuss, agree, disagree) for text pairs.
• Dataset was updated on 2021-12-27, providing a specific timestamp.

Limitations

• The dataset size, row count, and column details are unknown, limiting assessment of scale and feature richness.
• Data is from 2021, which may not reflect current news topics or linguistic patterns.

Provenance

Source

huggingface

Collection Method

Instances of (headline, body, stance) pairs, likely gathered for stance detection tasks.

Freshness

Last updated 2021-12-27.

Geography

Region tag suggests 'us', but specific coverage is unknown.