Unverified85% confidenceFactTime unknown
nmem的混合搜索将向量语义搜索和传统关键词搜索结合，通常采用Reciprocal Rank Fusion或线性加权的融合策略
1
Sources
85%
Confidence
Long-term
Relevance
6/1/2026
First Seen
Sources
nmem开源项目详解：六层记忆架构让AI Agent越用越聪明
githubdayyanj
Related Entities
Related Claims
UnverifiedContext Mode使用Reciprocal Rank Fusion算法合并两种搜索策略的排序结果77% similarVerified混合检索（Hybrid Search）将向量检索与BM25关键词检索的结果通过RRF（倒数排名融合）等算法合并，兼顾语义理解与精确匹配73% similarUnverified业界主流方案已演进为混合检索（Hybrid Search）策略，通过倒数排名融合（RRF）或加权线性组合合并向量检索与关键词检索结果，可提升召回率和精度72% similarVerified混合检索通常通过RRF（倒数排名融合，Reciprocal Rank Fusion）算法将两路检索结果加权合并70% similarUnverified混合检索（Hybrid Search）结合稠密向量语义相似性与BM25稀疏关键词匹配，两路结果经Re-Ranker模型重排后注入LLM上下文69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/31135API
curl https://kongchang.com/api/v1/knowledge/claims/31135MCP
get_claim(id=31135)