TypeRE Dataset

<b>TypeRE </b>is a supervised large-scale<b> relation extraction dataset</b> made from aligning public relation extraction datasets relations into a unified ontology. The public datasets used are: Wiki80, KBP37 and KnowledgeNet. The dataset includes a total of <i>30923 sentences</i> annotated with <i>28 relational</i> categories (including the no-relation category).<br><br><br>**TypeRE**是一款有监督的大规模**关系抽取数据集（relation extraction dataset）**，其构建方式为将多款公开关系抽取数据集的关系对齐至统一本体。所用公开数据集包括Wiki80、KBP37与KnowledgeNet。该数据集总计包含*30923条标注语句*，涵盖*28个关系类别*（包含无关系类别）。