RAG 企业应用实战

二、技术架构

核心组件

文档 → 切分 → 向量化 → 向量数据库
                    ↓
用户提问 → 向量化 → 相似度检索 → 上下文组装 → LLM → 回答

技术选型

| 组件 | 推荐方案 | |------|----------| | Embedding | OpenAI / BGE / M3E | | 向量数据库 | Chroma / FAISS / Milvus | | 框架 | LangChain / LlamaIndex | | LLM | GPT-4 / Claude / DeepSeek |

---

三、代码实现

安装依赖

pip install langchain langchain-openai chromadb

加载文档

from langchain.document_loaders import TextLoader, PDFLoader
# 加载文本文件
loader = TextLoader('company_policy.txt')
documents = loader.load()
# 加载 PDF
loader = PDFLoader('employee_handbook.pdf')
documents.extend(loader.load())

文档切分

from langchain.text_splitter import RecursiveCharacterTextSplitter
text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=500,
    chunk_overlap=50
)
chunks = text_splitter.split_documents(documents)
print(f"切分后共 {len(chunks)} 个文本块")

向量化存储

from langchain.embeddings import OpenAIEmbeddings
from langchain.vectorstores import Chroma
embeddings = OpenAIEmbeddings()
vectorstore = Chroma.from_documents(
    chunks,
    embeddings,
    persist_directory='./chroma_db'
)

检索问答

from langchain.chains import RetrievalQA
from langchain.chat_models import ChatOpenAI
llm = ChatOpenAI(model='gpt-4o-mini')
qa_chain = RetrievalQA.from_chain_type(
    llm=llm,
    chain_type='stuff',
    retriever=vectorstore.as_retriever(search_kwargs={'k': 3})
)
# 提问
question = "公司的年假政策是什么？"
answer = qa_chain.run(question)
print(answer)

---

四、优化技巧

1. 合理设置 Chunk 大小

from langchain.retrievers import ContextualCompressionRetriever
from langchain.retrievers.document_compressors import CohereRerank
compressor = CohereRerank()
compression_retriever = ContextualCompressionRetriever(
    base_compressor=compressor,
    base_retriever=vectorstore.as_retriever()
)

retriever = vectorstore.as_retriever(
    search_kwargs={
        'k': 5,
        'filter': {'department': 'HR'}
    }
)

from fastapi import FastAPI
app = FastAPI()
@app.post('/ask')
async def ask(question: str):
    answer = qa_chain.run(question)
    return {'answer': answer}

FROM python:3.11
WORKDIR /app
COPY . .
RUN pip install -r requirements.txt
CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000"]