知识卡片
文档级关系需要跨句证据
某关系可能要结合前一句的人名与后一句的描述才能判断,单句主谓宾规则不足以覆盖。选择训练与评价数据时应匹配跨句范围,并保留证据句及实体映射;文档级基准不是另一基准的简单重标版本。
这张卡片看不懂?
这张卡片已通过人工审查,不再接受失败反馈。
已有反馈 · 0 条
还没有人提交反馈。
在整书笔记中的位置
原书出处 · 1 处
- · 查看原书依据
The Document Relation Extraction Dataset (DocRED) expands the breadth of natural language text that can be used for relation extraction because it includes relations that require parsing multiple sentences of natural language text. The training and validation dataset used to train DocRED is one of the largest human-annotated datasets for document-level relation extraction. Most of the human-annotated knowledge graph data in DocRED is included in
参考来源
- 位置:《Natural Language Processing in Action, Second Edition》第11章《信息抽取与知识图谱》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Natural Language Processing in Action, Second Edition/_epub-src/OEBPS/Text/chapter-11.html)
- 结论依据:摘录中关于文档级关系需要跨句证据的定义、比较或条件支撑卡片;原书表述不严谨的范围已在覆盖记录逐项限定,不把案例结果外推为普遍结论。
- 原始内容:The Document Relation Extraction Dataset (DocRED) expands the breadth of natural language text that can be used for relation extraction because it includes relations that require parsing multiple sentences of natural language text. The training and validation dataset used to train DocRED is one of the largest human-annotated datasets for document-level relation extraction. Most of the human-annotated knowledge graph data in DocRED is included in