Knowledge-Enhanced Relation Extraction Dataset (KERED)

知识增强关系抽取数据集(KERED)是由北京理工大学医学技术学院的研究团队开发的，旨在通过结合证据句子和知识图谱来提升关系抽取的性能。KERED为每个句子标注了关系事实，并通过实体链接提供了实体的知识背景。该数据集包含三个子数据集：NYT10m、Wiki80和Wiki20m，总计约127万条实例。KERED的应用领域包括问答、知识图谱构建和阅读理解等，旨在解决传统文本关系抽取方法的局限性。

The Knowledge-Enhanced Relation Extraction Dataset (KERED) was developed by the research team from the School of Medical Technology, Beijing Institute of Technology. It is designed to improve the performance of relation extraction by integrating evidence sentences and knowledge graphs. KERED annotates relational facts for each sentence, and provides the knowledge background of entities through entity linking. This dataset comprises three subsets: NYT10m, Wiki80, and Wiki20m, with a total of approximately 1.27 million instances. The application fields of KERED include question answering, knowledge graph construction, reading comprehension and other scenarios, aiming to address the limitations of traditional text-based relation extraction methods.