未明确命名

本研究基于Stanford Question Answering Dataset (SQuADv1.1) 和 NewsQA，创建了一个用于隐式关系开放信息提取的大型数据集。该数据集包含107595条数据，主要来源于阅读理解任务，通过解析工具将问题答案对转换为开放信息提取数据集。数据集的创建过程涉及使用依赖解析器和15条规则来处理输入的问题答案对，输出隐式关系三元组。该数据集旨在提高开放信息提取系统对隐式关系的召回率，适用于自然语言理解等下游应用。

This study constructs a large-scale dataset for implicit relation open information extraction based on the Stanford Question Answering Dataset (SQuADv1.1) and NewsQA. The dataset contains 107,595 instances, which are mainly derived from reading comprehension tasks, and the question-answer pairs are converted into open information extraction datasets via parsing tools. The dataset construction process involves using a dependency parser and 15 rules to process the input question-answer pairs and output implicit relation triples. This dataset aims to improve the recall rate of implicit relations for open information extraction systems, and is applicable to downstream natural language understanding applications.