402,310 Spanish-language news title and body pairs extracted from the 2019 and 2020 Common Crawl CC-NEWS archives. The collection is partitioned into 370,125 training samples, 16,092 evaluation samples, and 16,092 test samples for natural language processing tasks.
Use Cases
- Train a sequence-to-sequence model to generate news titles from the body text
- Develop a text-classification system to categorize news articles based on the title and body content
- Perform sentiment-classification on Spanish news headlines
Strengths
- Contains 402,310 pairs of news titles and bodies
- Sourced from the 2019 and 2020 CC-NEWS Common Crawl data
- Includes dedicated splits for training (370,125), evaluation (16,092), and testing (16,092)
- Focuses exclusively on Spanish-language news content