[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"kb-rag-advanced":3},{"id":4,"title":5,"summary":6,"source":7,"category":8,"date":9,"url":10,"content":11},"rag-advanced","高级 RAG 工程实践：从朴素检索到混合检索","结合稀疏检索与稠密检索，通过查询改写、路由与重排序，把 RAG 精度提升一个量级。","InfoQ 中文","RAG","2026-08-10","https:\u002F\u002Fwww.infoq.cn\u002F",[12,13,14,15],"朴素 RAG（向量检索 + 拼接提示）在复杂问题上表现不稳定，工程化改造势在必行。","混合检索：BM25 稀疏检索负责精确匹配，向量检索负责语义召回，融合排序可兼顾两者优势。","查询改写（Query Rewriting）在多轮对话中尤为关键，可将口语化问题改写为可检索的表达。","重排序（Rerank）模型虽然带来额外延迟，但对精度提升显著，是生产环境的标配组件。"]