A Chinese text classification dataset sourced from the Toutiao (Today's Headlines) client. The dataset contains news entries where each row includes a news ID, category code, category name, headline string, and keywords, with fields separated by '_!_'. The dataset was uploaded to Hugging Face by author fourteenBDr and last updated on 2022-08-21.
Use Cases
- Train a multiclass text classifier based on the provided news category labels.
- Perform keyword extraction or analysis based on the provided news keywords field.
- Conduct topic modeling on Chinese news headlines based on the headline text and category structure.
- Benchmark model performance on a real-world Chinese news categorization task.
Strengths
- Data includes explicit category codes and names for supervised learning.
- Each entry contains a headline and a separate keywords field, providing multiple text features.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Last updated 2022-08-21 14:58:22; freshness should be verified.
- Row count is unknown, which may limit suitability assessment.
Provenance
- Source
- Toutiao (Today's Headlines) client
- Collection Method
- Scraped or collected from the news platform client.
- Time Range
- null
- Freshness
- 2022-08-21
- Geography
- null