<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title><![CDATA[RAG 进阶：混合检索（Hybrid Search）提升准确率 40%]]></title><description><![CDATA[<blockquote>
<p dir="auto">来源：AI 订阅指南</p>
</blockquote>
<p dir="auto">纯向量检索有时不够，混合检索能显著提升效果。</p>
<p dir="auto"><strong>什么是混合检索？</strong><br />
结合关键词检索（BM25）和向量检索，取两者之长。</p>
<p dir="auto"><strong>实现方式：</strong></p>
<pre><code class="language-python">from langchain.retrievers import EnsembleRetriever

# BM25 检索器（关键词）
bm25 = BM25Retriever.from_documents(docs)

# 向量检索器
vector = db.as_retriever()

# 混合检索
ensemble = EnsembleRetriever(
    retrievers=[bm25, vector],
    weights=[0.3, 0.7]  # BM25 30%，向量 70%
)
</code></pre>
<p dir="auto"><strong>效果</strong>：</p>
<ul>
<li>纯向量检索：准确率 ~70%</li>
<li>混合检索：准确率 ~85%</li>
</ul>
<p dir="auto"><strong>适合场景</strong>：专业术语多的领域（医疗、法律、技术文档）。</p>
<hr />
<p dir="auto"><em>更多 RAG 教程请关注 AI 订阅指南。</em></p>
]]></description><link>https://aspxai.com/topic/233/rag-进阶-混合检索-hybrid-search-提升准确率-40</link><generator>RSS for Node</generator><lastBuildDate>Mon, 22 Jun 2026 07:55:51 GMT</lastBuildDate><atom:link href="https://aspxai.com/topic/233.rss" rel="self" type="application/rss+xml"/><pubDate>Mon, 22 Jun 2026 03:02:00 GMT</pubDate><ttl>60</ttl><item><title><![CDATA[Reply to RAG 进阶：混合检索（Hybrid Search）提升准确率 40% on Mon, 22 Jun 2026 03:03:33 GMT]]></title><description><![CDATA[<p dir="auto">微调一个 7B 模型大概多少钱？有没有便宜的方案？</p>
]]></description><link>https://aspxai.com/post/1082</link><guid isPermaLink="true">https://aspxai.com/post/1082</guid><dc:creator><![CDATA[松风长安]]></dc:creator><pubDate>Mon, 22 Jun 2026 03:03:33 GMT</pubDate></item><item><title><![CDATA[Reply to RAG 进阶：混合检索（Hybrid Search）提升准确率 40% on Mon, 22 Jun 2026 03:03:33 GMT]]></title><description><![CDATA[<p dir="auto">ChromaDB 在小数据量下够用，数据量大了建议换 Milvus。</p>
]]></description><link>https://aspxai.com/post/1081</link><guid isPermaLink="true">https://aspxai.com/post/1081</guid><dc:creator><![CDATA[云端寻路]]></dc:creator><pubDate>Mon, 22 Jun 2026 03:03:33 GMT</pubDate></item></channel></rss>