text2sql

text2sql数据集--spider Spider数据集是耶鲁大学于2018年新提出的一个较大规模的NL2SQL数据集。该数据集包含了10,181条自然语言问句，分布在200个独立数据库中的5,693条SQL，内容覆盖了138个不同的领域。虽然在数据数量上不如WikiSQL，但Spider引入了更多的SQL用法，例如Group By、Order By、Having等高阶操作，甚至需要Join不同表，更贴近真实场景，所以难度也更大。text2SQL Dataset -- Spider: The Spider dataset is a large-scale NL2SQL dataset proposed by Yale University in 2018. This dataset contains 10,181 natural language questions, along with 5,693 SQL queries across 200 independent databases, covering 138 distinct domains. Although its overall scale is smaller than that of WikiSQL, Spider incorporates more advanced SQL usages such as Group By, Order By, Having and other high-level operations, and even requires joining different tables, making it closer to real-world scenarios and thus more challenging.

搜集汇总

背景与挑战

背景概述

text2sql数据集基于耶鲁大学2018年提出的Spider数据集，包含10,181条自然语言问句和5,693条SQL查询，覆盖200个独立数据库和138个领域。该数据集引入了高阶SQL操作如Group By、Order By、Having和Join，更贴近真实应用场景，具有较高难度。