TableQA

TableQA是由追一科技创建的大规模中文NL2SQL数据集，包含64,891个问题和20,311个独特的SQL查询，覆盖超过6,000个来自不同领域的表格。数据集的创建过程中，注释者被鼓励提出与表格中存储的实体或术语不同的问题，并设计了一些无法根据给定表格回答的问题，以更贴近实际场景。TableQA主要用于解决自然语言到结构化查询语言的转换问题，特别强调实体链接和可回答性，为NL2SQL研究提供了新的挑战和机遇。TableQA is a large-scale Chinese NL2SQL dataset developed by Zhuiyi Technology. It contains 64,891 questions and 20,311 unique SQL queries, covering over 6,000 tables from diverse domains. During the dataset construction process, annotators were encouraged to formulate questions that differ from the entities or terms stored in the tables, and some questions unanswerable based on the given tables were designed to better align with real-world scenarios. TableQA is primarily used to address the task of converting natural language to structured query language, with a particular emphasis on entity linking and answerability, providing new challenges and opportunities for NL2SQL research.

搜集汇总

背景与挑战

背景概述

TableQA是一个由追一科技创建的大规模中文NL2SQL数据集，包含64,891个问题和20,311个独特的SQL查询，覆盖超过6,000个来自不同领域的表格。它强调实体链接和可回答性，通过设计无法回答的问题来贴近实际场景，主要用于自然语言到结构化查询语言的转换研究，为NL2SQL领域提供新的挑战和机遇。