WebQueryTable

该数据集名为WebQueryTable，包含了从一家商业搜索引擎的搜索日志中收集的21,113条查询，以及273,816张表格。经过人工评估后，每个查询都获得了一张相关的表格。此外，还提供了表格的标题作为上下文信息。该数据集采用了最初发布的训练集、验证集和测试集的划分方式。其规模涉及273,816张表格和21,113条查询，针对的任务是自然语言表格检索（Nltr）。

This dataset is named WebQueryTable. It contains 21,113 queries and 273,816 tables collected from the search logs of a commercial search engine. After manual evaluation, each query is paired with a relevant table. Additionally, the titles of the tables are provided as contextual information. The dataset adopts the original training, validation, and test set splits released in the initial publication. With a scale of 273,816 tables and 21,113 queries, this dataset is designed for the task of Natural Language Table Retrieval (NLTR).