News Article Dataset for Moderation Testing

A dataset for testing news classification or moderation models, created by myradeng and uploaded to Hugging Face in May 2022. It contains text data categorized by a 'moderate' label, likely for content filtering tasks. The dataset size is estimated between 10,000 and 100,000 rows based on platform tags.

Use Cases

• Train a text classifier to predict the 'moderate' label from news article content.
• Benchmark model performance on a text-based news moderation task using the provided labels.
• Analyze linguistic patterns in news articles flagged for moderation.
• Fine-tune a language model for content safety using the labeled news text.

Strengths

• Dataset size is categorized as between 10,000 and 100,000 rows.
• Data modality includes structured (tabular) and unstructured (text) formats.
• Provides a specific label ('moderate') for supervised learning tasks.

Limitations

• Exact row count, column names, and data schema are unknown.
• Data is from 2022 or earlier, potentially missing recent news trends.
• Geographic coverage is indicated as US-only, limiting global applicability.

Provenance

Source

Hugging Face user myradeng.

Collection Method

Unknown collection or generation method.

Time Range

null

Freshness

Last updated in May 2022; static dataset with no stated update frequency.

Geography

United States (inferred from 'Regionus' platform tag).