Unverified90% confidenceFactTime unknown
OpenClaw的记忆检索本质上是RAG（检索增强生成），将记忆文件切成Chunk，通过字面匹配和语义Embedding的加权相似度排序，取Top-K结果返回给语言模型
1
Sources
90%
Confidence
Long-term
Relevance
6/3/2026
First Seen
Sources
Related Entities
Related Claims
VerifiedModern RAG implementations widely incorporate dense retrieval (Dense Retrieval), using embedding models to transform text into high-dimensional vectors and performing matching via semantic similarity.76% similarUnverifiedCodex adopts a RAG (Retrieval-Augmented Generation) approach, combining semantic similarity and code dependency graphs for streamlined retrieval, extracting only code snippets relevant to the current task.68% similarUnverified生成式引擎通常采用检索增强生成（RAG）架构，先从索引语料库检索相关片段，再由语言模型综合生成自然语言答案66% similarUnverifiedSolutions for semantic search via lyrics text rely on a combination of two technologies: vector embeddings and retrieval-augmented generation (RAG).66% similarUnverifiedMEMOIR uses Semantic Paths instead of flat UUID indexes to organize memories, reducing retrieval complexity from O(n) to O(log n).66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/40386API
curl https://kongchang.com/api/v1/knowledge/claims/40386MCP
get_claim(id=40386)