如何结合langchain、neo4j实现关联检索问答

liliangcsdn

已于 2025-10-23 09:12:35 修改

阅读量529

点赞数 14

CC 4.0 BY-SA版权

文章标签： python 开发语言 langchain neo4j

于 2025-10-23 09:10:46 首次发布

本文链接：https://blog.csdn.net/liliang199/article/details/153687193

langchain时llm应用最流行的开发工具之一，neo4j时应用最广泛的图数据库管理工具。

这里尝试结合langchain和neo4j，示例关联检索的实现过程。

以下内容中的测试例和代码，整理和修改自网络资料。

1 工具安装

这里安装neo4j、langchain工具包等必要工具。

1.1 neo4j安装

为简化操作这里采用docker安装neo4j，并假设docker和neo4j已经安装。

安装过程参考参考如下链接

neo4j安装

https://blog.csdn.net/liliang199/article/details/153691513

apoc安装

https://blog.csdn.net/liliang199/article/details/153693401

1.2 langchain包安装

基于conda构建langchain的测试环境，安装neo4j相关依赖包。

包括cmake、ninja、rust、gcc/g++、tiktoken等。

conda create -n langchain python=3.10

conda activate langchain

conda install cmake

conda install ninja

conda install -c conda-forge gcc=12 gxx=12

conda install conda-forge::rust

pip install tiktoken -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install py2neo -i https://pypi.tuna.tsinghua.edu.cn/simple

pip install langchain-neo4j -i https://pypi.tuna.tsinghua.edu.cn/simple

pip install langchain langchain-openai langgraph -i https://pypi.tuna.tsinghua.edu.cn/simple

2.3 ollama安装

建议安装如下版本，避免版本之间的冲突。

pip install langchain-core==0.3.78 -i https://pypi.tuna.tsinghua.edu.cn/simple
pip install langchain_ollama==0.3.10 -i https://pypi.tuna.tsinghua.edu.cn/simple

2 数据准备

2.1 设置neo4j连接

neo4j连接信息以环境变量的方式设置，代码示例如下。

import os

os.environ["NEO4J_URI"] = "bolt://localhost:7687"
os.environ["NEO4J_USERNAME"] = "neo4j"
os.environ["NEO4J_PASSWORD"] = "12345678"

2.2 导入数据

测试数据来源于blog-datasets的movies_small.csv，连接如下。

https://raw.githubusercontent.com/tomasonjo/blog-datasets/main/movies/movies_small.csv

由于github访问受限，下载movies_small.csv后使用minso创建一个本地的下载链接，假设为

http://host_ip:9000/tomasonjo.blog-datasets/movies/movies_small.csv

minio和neo4j是docker分别部署的，其中ip需要设置为宿主机ip，因为neo4j容器不能直接访问minio容器的链接。

from langchain_neo4j import Neo4jGraph

graph = Neo4jGraph()

# Import movie information

movies_query = """
LOAD CSV WITH HEADERS FROM 
'http://host_ip:9000/tomasonjo.blog-datasets/movies/movies_small.csv'
AS row
MERGE (m:Movie {id:row.movieId})
SET m.released = date(row.released),
    m.title = row.title,
    m.imdbRating = toFloat(row.imdbRating)
FOREACH (director in split(row.director, '|') | 
    MERGE (p:Person {name:trim(director)})
    MERGE (p)-[:DIRECTED]->(m))
FOREACH (actor in split(row.actors, '|') | 
    MERGE (p:Person {name:trim(actor)})
    MERGE (p)-[:ACTED_IN]->(m))
FOREACH (genre in split(row.genres, '|') | 
    MERGE (g:Genre {name:trim(genre)})
    MERGE (m)-[:IN_GENRE]->(g))
"""

graph.query(movies_query)

3 功能验证

这里示例基于langchain+neo4j的基础关联检索。

3.1 展示图schema

通过以下代码，输出图graph schema。

graph.refresh_schema()
print(graph.schema)

graph schema如下所示

Node properties:
Person {name: STRING, age: INTEGER}
Movie {imdbRating: FLOAT, id: STRING, released: DATE, title: STRING}
Genre {name: STRING}
Relationship properties:

The relationships:
(:Person)-[:knows]->(:Person)
(:Person)-[:DIRECTED]->(:Movie)
(:Person)-[:ACTED_IN]->(:Movie)
(:Movie)-[:IN_GENRE]->(:Genre)

将enhanced_schema设置为True，表示输出propertiy，获取更多的scheme信息。

enhanced_graph = Neo4jGraph(enhanced_schema=True)
print(enhanced_graph.schema)

输出如下

Node properties:
- **Person**
- `name`: STRING Example: "张三"
- `age`: INTEGER Min: 30, Max: 30
- **Movie**
- `imdbRating`: FLOAT Min: 2.4, Max: 9.3
- `id`: STRING Example: "1"
- `released`: DATE Min: 1964-12-16, Max: 1996-09-15
- `title`: STRING Example: "Toy Story"
- **Genre**
- `name`: STRING Example: "Adventure"
Relationship properties:

The relationships:
(:Person)-[:knows]->(:Person)
(:Person)-[:DIRECTED]->(:Movie)
(:Person)-[:ACTED_IN]->(:Movie)
(:Movie)-[:IN_GENRE]->(:Genre)

3.2 GraphQACypherChain

langchain结合graph图，回答用户问题的流程如下所示。

GraphQACypherChain接收用户问题，使用llm转化为cypher、运行cypher查询、将查询结果使用llm转化为对用户问题的回答。

以下是示例代码

import os
os.environ['OPENAI_API_KEY'] = "sk-xxxx"
os.environ['OPENAI_BASE_URL'] = "https://llm_provider/v1"

from langchain_neo4j import GraphCypherQAChain
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(model="deepseek-v3", temperature=0)
chain = GraphCypherQAChain.from_llm(
    graph=enhanced_graph, llm=llm, verbose=True, allow_dangerous_requests=True
)
response = chain.invoke({"query": "What was the cast of the Casino?"})
response

输出如下

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (p:Person)-[:ACTED_IN]->(m:Movie {title: "Casino"})
RETURN p.name AS cast_member
Full Context:
[{'cast_member': 'Joe Pesci'}, {'cast_member': 'Robert De Niro'}, {'cast_member': 'Sharon Stone'}, {'cast_member': 'James Woods'}]

> Finished chain.
{'query': 'What was the cast of the Casino?',
'result': 'The cast of *Casino* includes Joe Pesci, Robert De Niro, Sharon Stone, and James Woods.'}