Unverified50% confidenceFactExact time
RAG架构核心思路是将检索与生成两个模块解耦，检索模块从外部知识库实时拉取相关内容，生成模块以检索结果作为锚点组织输出
1
Sources
50%
Confidence
Long-term
Relevance
7/12/2026
First Seen
Sources
Related Claims
VerifiedRAG（检索增强生成）是一种结合信息检索与生成式大模型的架构范式，模型生成答案前先根据用户查询从外部知识库检索最相关的文档片段，再将其作为上下文注入模型82% similarVerifiedThe core architecture of RAG is divided into three key stages: document preprocessing and vectorization, semantic retrieval, and augmented generation.82% similarVerifiedRAG（检索增强生成）技术架构将文档切片、向量化、存入向量数据库，查询时先检索相关片段再交给模型生成回答81% similarUnverifiedRAG's core architecture consists of two stages: retrieval and generation80% similarUnverifiedRAG typically consists of three core stages: offline indexing, online retrieval, and context injection.78% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/493431API
curl https://kongchang.com/api/v1/knowledge/claims/493431MCP
get_claim(id=493431)