知识卡片
实体模型架构应查看实际配置
不同NER系统可使用CNN、循环结构或Transformer,但不能据此把某框架默认实现说成任意常见架构。spaCy EntityRecognizer是转移式跨度识别器,实际表示与组件依配置;应核对所载版本,不把通用技术列表当实现说明。
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
So you can see how the context of the entity—both the words that came before it and after it, even if they are far away in the sentence—matters. That’s why the popular approaches to NER with neural networks include multilevel CNNs, bidirectional transformers, such as BERT, or bidirectional LSTMs. The last one, combined with a technique called conditional random fields (CRFs), is what spaCy uses
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:摘录中关于实体模型架构应查看实际配置的定义、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:So you can see how the context of the entity—both the words that came before it and after it, even if they are far away in the sentence—matters. That’s why the popular approaches to NER with neural networks include multilevel CNNs, bidirectional transformers, such as BERT, or bidirectional LSTMs. The last one, combined with a technique called conditional random fields (CRFs), is what spaCy uses