DuIE2.0中文关系抽取数据集

DuIE2.0是业界规模最大的中文关系抽取数据集，其schema在传统简单关系类型基础上添加了多元复杂关系类型，此外其构建语料来自百度百科、百度信息流及百度贴吧文本，全面覆盖书面化表达及口语化表达语料，能充分考察真实业务场景下的关系抽取能力。

DuIE 2.0 is the largest Chinese relation extraction dataset in the industry. Its schema incorporates multiple complex relation types on the basis of traditional simple relation categories. The corpus of this dataset is collected from texts of Baidu Encyclopedia, Baidu Information Flow and Baidu Tieba, which comprehensively covers both formal written and colloquial expressions. This allows for a thorough evaluation of relation extraction capabilities in real-world business scenarios.