127,600 news articles categorized into four distinct classes: World, Sports, Business, and Science/Technology. Each entry includes a headline and a short summary of the news story, providing a balanced distribution for text classification tasks.
Use Cases
- Train multi-class classification models to map 'Title' and 'Description' text to one of four category labels
- Benchmark short-text embedding techniques using the concise 'Description' field
- Develop news aggregation filters by identifying 'World' or 'Business' topics within raw text strings
Strengths
- 127,600 total records split into 120,000 training and 7,600 testing samples
- Four mutually exclusive labels: World (1), Sports (2), Business (3), and Sci/Tech (4)
- Structured with 'Title' and 'Description' fields for each news item