知识卡片
解析质量决定可检索证据
提取正文、标题层级、表格和元数据是后续分块与检索的基础。若解析丢失单位、列标题或否定,检索器再准确也无法返回正确证据;应先检查抽取结果及源位置,而非只优化嵌入模型。
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
- · 表示学习与嵌入
原书出处 · 1 处
- · 查看原书依据
Document Parsing Unstructured data first needs to be processed to make it amenable to retrieval. This usually involves parsing text from the document, splitting it into manageable units, associating metadata with these units, generating embeddings, storing, and indexing them for easy access. If it makes sense for your use case to have sentences as the basic unit of text, NLTK’s Punkt tokenizer is a tried
参考来源
- 位置:《Designing Large Language Model Applications》第11章《表示学习与嵌入》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch11.html)
- 结论依据:摘录中关于解析质量决定可检索证据的定义、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:Document Parsing Unstructured data first needs to be processed to make it amenable to retrieval. This usually involves parsing text from the document, splitting it into manageable units, associating metadata with these units, generating embeddings, storing, and indexing them for easy access. If it makes sense for your use case to have sentences as the basic unit of text, NLTK’s Punkt tokenizer is a tried