SQA

该数据集包含了6,066个独特的问答序列，共17,553个问题-答案对，旨在将复杂问题分解为更简单且相互关联的连续性问题。这些问题被分类为列选择、子集选择和行选择等类别。数据规模方面，拥有6,066个独特的问答序列和17,553个问题-答案对，任务类型为顺序性问题解答。This dataset contains 6,066 unique question-answer sequences, with a total of 17,553 question-answer pairs. It is designed to decompose complex questions into simpler, interrelated sequential questions. The questions are categorized into types such as column selection, subset selection, and row selection, and the task of this dataset is sequential question answering.

搜集汇总

背景与挑战

背景概述

WikiTableQuestions（可能简称SQA）是一个用于半结构化HTML表格问答任务的数据集，包含22033个示例，分为训练集和测试集，专注于复杂问题并基于维基百科表格。数据集提供多格式数据（如TSV、CSV）和CoreNLP标注，支持语义解析研究，并附有官方评估工具。