DACO

该数据集名为DACO，包含了来自现实场景的440个数据库，大约2664个查询-答案对用于弱监督，以及一个经过人工精细标注的高质量测试集。该数据集不仅包括自动标注的查询-答案对，还包含了经过人工精炼的版本，特别关注于多样化的应用驱动型查询。其规模涵盖了440个数据库和大约2000个查询-答案对，任务重点在于数据分析和查询解答。

This dataset is named DACO. It encompasses 440 real-world databases, approximately 2,664 query-answer pairs for weak supervision, and a high-quality manually fine-annotated test set. In addition to the automatically annotated query-answer pairs, it also includes a manually refined version with a particular focus on diverse application-driven queries. The dataset covers 440 databases and around 2,000 query-answer pairs, with the core task focusing on data analysis and query answering.