Unverified85% confidenceFactTime unknown
传统RAG无法回答'知识库里有哪些文档'这类元数据问题，因为它本质上只能检索内容片段
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe core limitations of traditional RAG solutions include: document chunking destroying the holistic semantic structure of knowledge, vector similarity retrieval failing to capture deep logical associations, and each query being independent with no ability to accumulate structured knowledge.72% similarUnverified大多数RAG（检索增强生成）实现只解决了'检索到什么'的问题，却忽视了检索到的内容是否可信、是否互相矛盾的更深层问题71% similarUnverifiedTraditional RAG systems have three major limitations: returning empty results when the answer cannot be retrieved, inability to answer metadata questions, and lack of multi-turn iterative error correction.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/27096API
curl https://kongchang.com/api/v1/knowledge/claims/27096MCP
get_claim(id=27096)