LangChain-v0.2文档翻译：2.8、教程-构建查询分析系统

介绍
教程
2.1. 构建一个简单的 LLM 应用程序
2.2. 构建一个聊天机器人
2.3. 构建向量存储库和检索器
2.4. 构建一个代理
2.5. 构建检索增强生成 (RAG) 应用程序
2.6. 构建一个会话式RAG应用程序
2.7. 在SQL数据上构建一个问答系统
2.8. 构建查询分析系统（点击查看原文）

**构建查询分析系统

将展示如何在一个基本的端到端示例中使用查询分析。这将涵盖创建一个简单的搜索引擎，展示当将原始用户问题传递给该搜索时发生的故障模式，然后是一个查询分析如何帮助解决这个问题的例子。有许多不同的查询分析技术，这个端到端的示例不会展示所有的技术。

为了这个示例的目的，我们将在LangChain YouTube视频上进行检索。

设置

安装依赖

# 安装依赖库
# pip install -qU langchain langchain-community langchain-openai youtube-transcript-api pytube langchain-chroma

设置环境变量

我们将在这个示例中使用OpenAI：

import getpass
import os

os.environ["OPENAI_API_KEY"] = getpass.getpass()

# 可选，取消注释以使用LangSmith追踪运行。在这里注册：https://smith.langchain.com。
# os.environ["LANGCHAIN_TRACING_V2"] = "true"
# os.environ["LANGCHAIN_API_KEY"] = getpass.getpass()

加载文档

我们可以使用YouTubeLoader来加载一些LangChain视频的字幕：

from langchain_community.document_loaders import YoutubeLoader

urls = [
    # 这里是一些YouTube视频的URL列表
]

docs = []
for url in urls:
    docs.extend(YoutubeLoader.from_youtube_url(url, add_video_info=True).load())

import datetime

# 添加一些额外的元数据：视频发布的年份
for doc in docs:
    doc.metadata["publish_year"] = int(
        datetime.datetime.strptime(
            doc.metadata["publish_date"], "%Y-%m-%d %H:%M:%S"
        ).strftime("%Y")
    )

以下是我们已加载视频的标题：

[doc.metadata["title"] for doc in docs]

['OpenGPTs',
 'Building a web RAG chatbot: using LangChain, Exa (prev. Metaphor), LangSmith, and Hosted Langserve',
 'Streaming Events: Introducing a new `stream_events` method',
 'LangGraph: Multi-Agent Workflows',
 'Build and Deploy a RAG app with Pinecone Serverless',
 'Auto-Prompt Builder (with Hosted LangServe)',
 'Build a Full Stack RAG App With TypeScript',
 'Getting Started with Multi-Modal LLMs',
 'SQL Research Assistant',
 'Skeleton-of-Thought: Building a New Template from Scratch',
 'Benchmarking RAG over LangChain Docs',
 'Building a Research Assistant from Scratch',
 'LangServe and LangChain Templates Webinar']

这里是与每个视频相关联的元数据。我们可以看到每个文档还具有标题、浏览次数、发布日期和长度：

docs[0].metadata

{
    'source': 'HAn9vnJy6S4',
    'title': 'OpenGPTs',
    'description': 'Unknown',
    'view_count': 7210,
    'thumbnail_url': 'https://i.ytimg.com/vi/HAn9vnJy6S4/hq720.jpg',
    'publish_date': '2024-01-31 00:00:00',
    'length': 1530,
    'author': 'LangChain',
    'publish_year': 2024
}

这是文档内容的一个样本：

docs[0].page_content[:500]

"hello today I want to talk about open gpts open gpts is a project that we built here at linkchain uh that replicates the GPT store in a few ways so it creates uh end user-facing friendly interface to create different Bots and these Bots can have access to different tools and they can uh be given files to retrieve things over and basically it's a way to create a variety of bots and expose the configuration of these Bots to end users it's all open source um it can be used with open AI it can be us"

索引文档

每当我们执行检索时，我们需要创建一个我们可以查询的文档索引。我们将使用向量存储来索引我们的文档，我们首先将它们分块，以使我们的检索更加简洁和精确：

from langchain_chroma import Chroma
from langchain_openai import OpenAIEmbeddings
from langchain_text_splitters import RecursiveCharacterTextSplitter

text_splitter = RecursiveCharacterTextSplitter(chunk_size=2000)
chunked_docs = text_splitter.split_documents(docs)
embeddings = OpenAIEmbeddings(model="text-embedding-3-small")
vectorstore = Chroma.from_documents(
    chunked_docs,
    embeddings,
)

没有查询分析的检索

我们可以直接对用户问题执行相似性搜索，以找到与问题相关的片段：

search_results = vectorstore.similarity_search("how do I build a RAG agent")
print(search_results[0].metadata["title"])
print(search_results[0].page_content[:500])

Build and Deploy a RAG app with Pinecone Serverless
hi this is Lance from the Lang chain team and today we're going to be building and deploying a rag app using pine con serval list from scratch so we're going to kind of walk through all the code required to do this and I'll use these slides as kind of a guide to kind of lay the the ground work um so first what is rag so under capoy has this pretty nice visualization that shows LMS as a kernel of a new kind of operating system and of course one of the core components of our operating system is th

效果非常好！我们的第一个结果与问题非常相关。

如果我们想要搜索特定时间段的结果会怎样？

search_results = vectorstore.similarity_search("videos on RAG published in 2023")
print(search_results[0].metadata["title"])
print(search_results[0].metadata["publish_date"])
print(search_results[0].page_content[:500])

OpenGPTs
2024-01-31
hardcoded that it will always do a retrieval step here the assistant decides whether to do a retrieval step or not sometimes this is good sometimes this is bad sometimes it you don't need to do a retrieval step when I said hi it didn't need to call it tool um but other times you know the the llm might mess up and not realize that it needs to do a retrieval step and so the rag bot will always do a retrieval step so it's more focused there because this is also a simpler architecture so it's always

我们的第一个结果是2024年的（尽管我们要求的是2023年的视频），并且与输入不太相关。由于我们只是针对文档内容进行搜索，结果无法根据任何文档属性进行过滤。

这只是可能出现的一种失败模式。现在让我们来看看一个基本的查询分析如何可以修复它！

查询分析

我们可以使用查询分析来改进检索结果。这将涉及定义一个包含一些日期过滤器的查询模式，并使用函数调用模型将用户问题转换为结构化查询。

查询模式

在这种情况下，我们将有明确的发布日期的最小和最大属性，以便可以对其进行过滤。

from typing import Optional

from langchain_core.pydantic_v1 import BaseModel, Field

class Search(BaseModel):
    """Search over a database of tutorial videos about a software library."""
    query: str = Field(
        ...,
        description="Similarity search query applied to video transcripts.",
    )
    publish_year: Optional[int] = Field(None, description="Year video was published")

查询生成

为了将用户问题转换为结构化查询，我们将利用OpenAI的工具调用API。具体来说，我们将使用新的ChatModel.with_structured_output()构造函数来处理将模式传递给模型并解析输出。

from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_openai import ChatOpenAI

system = """
You are an expert at converting user questions into database queries.
You have access to a database of tutorial videos about a software library for building LLM-powered applications.
Given a question, return a list of database queries optimized to retrieve the most relevant results.

If there are acronyms or words you are not familiar with, do not try to rephrase them.
"""

prompt = ChatPromptTemplate.from_messages(
    [
        ("system", system),
        ("human", "{question}"),
    ]
)
llm = ChatOpenAI(model="gpt-3.5-turbo-0125", temperature=0)
structured_llm = llm.with_structured_output(Search)
query_analyzer = {"question": RunnablePassthrough()} | prompt | structured_llm

/Users/bagatur/langchain/libs/core/langchain_core/_api/beta_decorator.py:86: LangChainBetaWarning: The function `with_structured_output` is in beta. It is actively being worked on, so the API may change.
  warn_beta(

让我们看看我们的分析器为我们之前搜索的问题生成了哪些查询：

query_analyzer.invoke("how do I build a RAG agent")

Search(query='build RAG agent', publish_year=None)

query_analyzer.invoke("videos on RAG published in 2023")

Search(query='RAG', publish_year=2023)

带查询分析的检索

我们的查询分析看起来相当不错；现在让我们尝试使用我们生成的查询实际执行检索。

**注意：**在我们的示例中，我们指定了tool_choice="Search"。这将强制LLM调用一个 - 且只有一个 - 工具，这意味着我们总是有一个优化的查询可以查找。请注意，情况并非总是如此 - 请参阅其他指南，了解在没有 - 或多个 - 优化查询返回时如何处理。

from typing import List

from langchain_core.documents import Document

def retrieval(search: Search) -> List[Document]:
    if search.publish_year is not None:
        # 这是特定于Chroma的语法，
        # 我们正在使用的向量数据库。
        _filter = {"publish_year": {"$eq": search.publish_year}}
    else:
        _filter = None
    return vectorstore.similarity_search(search.query, filter=_filter)

retrieval_chain = query_analyzer | retrieval

现在我们可以在之前有问题的输入上运行这个链，看到它只产生那一年的结果！

results = retrieval_chain.invoke("RAG tutorial published in 2023")

[(doc.metadata["title"], doc.metadata["publish_date"]) for doc in results]

[
    ('Getting Started with Multi-Modal LLMs', '2023-12-20 00:00:00'),
    ('LangServe and LangChain Templates Webinar', '2023-11-02 00:00:00'),
    ('Getting Started with Multi-Modal LLMs', '2023-12-20 00:00:00'),
    ('Building a Research Assistant from Scratch', '2023-11-16 00:00:00')
]