SAOKE

该数据集包含了由众包方式标注的超过四万句句子及其对应的SAOKE格式的事实，专门用于开放信息提取任务。此外，SAOKE数据集是目前公开可用的最大的人力标注数据集，用于开放信息提取任务，其句子和事实涵盖了广泛的不同领域。该数据集规模宏大，包含超过四万句句子和大约十万条事实。其任务目标是开放信息提取。

This dataset, which contains over 40,000 crowdsourced annotated sentences and their corresponding SAOKE-format facts, is specifically developed for open information extraction (OIE) tasks. Furthermore, the SAOKE dataset is currently the largest publicly available human-annotated dataset for open information extraction, with its sentences and facts spanning a wide range of domains. This large-scale dataset includes more than 40,000 sentences and approximately 100,000 facts, with its core task being open information extraction.

搜集汇总

背景与挑战

背景概述

SAOKE是一个开放领域信息抽取（OIE）数据集，旨在从开放领域句子中提取关键信息，支持人工智能应用如文本摘要。数据集包含一个16.80MB的JSON文件，发布于2022年11月，目前处于未迁移至新版本的临时状态。