A collection of NBA game statistics paired with human-written summaries for data-to-text generation. It aligns structured box score data with narrative descriptions of professional basketball games.
Use Cases
- Train models to generate narrative summaries from NBA game stats.
- Evaluate the factual consistency of generated text against the provided human-written summaries.
- Map entities in the human written summaries back to the structured NBA game stats.
Strengths
- Includes NBA game statistics as the structured data source.
- Contains human-written summaries aligned with game data.
- Designed for data-to-text generation research.