知识卡片
图检索用关系支持全局主题
从语料抽取实体关系并形成层级社区和摘要,可帮助回答跨文档连接或全局主题问题。建图成本高,遗漏与伪关系会传入摘要;图结构不是天然正确知识,也不是所有全局问答必需。
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
- · 检索增强生成
原书出处 · 1 处
- · 查看原书依据
Finally, let’s discuss GraphRAG, an up-and-coming retrieval paradigm that leverages knowledge graphs for better retrieval. GraphRAG A key limitation of the retrieval approaches we have discussed so far is their inability to facilitate answering questions that require drawing connections between different parts of the document corpus, as well as questions that involve summarizing high-level themes across the dataset. For example, all the retrieval techniques we
参考来源
- 位置:《Designing Large Language Model Applications》第12章《检索增强生成》(来源项目:a19c7933-4eda-4176-aebd-30fe7e1ad292;源文件:extracted/Designing Large Language Model Applications/_epub-src/OEBPS/ch12.html)
- 结论依据:摘录中关于图检索用关系支持全局主题的定义、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:Finally, let’s discuss GraphRAG, an up-and-coming retrieval paradigm that leverages knowledge graphs for better retrieval. GraphRAG A key limitation of the retrieval approaches we have discussed so far is their inability to facilitate answering questions that require drawing connections between different parts of the document corpus, as well as questions that involve summarizing high-level themes across the dataset. For example, all the retrieval techniques we