langchain+neo4j 实现图谱GraphRAG

整体思路参考如下

1.首先单独conda create 一个环境，用来安装依赖

pip install --upgrade --quiet  langchain langchain-community langchain-openai langchain-experimental neo4j

2.首先是基础代码，这一步先查看当前模型的能力是否能够提取出文本中的实体和关系。（在这之前，笔者尝试了deepseek,glm4等api）其中glm-4 api10次大概有2次能够识别正确结果。
后切换为qwen api，初步代码如下，参考的是langchain + neo4j

#!/usr/bin/python
# -*- coding: <utf-8> -*-
from langchain.docstore.document import Document
from langchain.text_splitter import CharacterTextSplitter
from langchain_community.document_loaders import TextLoader
from langchain_community.vectorstores import Neo4jVector
from langchain_openai import OpenAIEmbeddings, ChatOpenAI
from langchain_community.graphs import Neo4jGraph
from langchain_experimental.graph_transformers import LLMGraphTransformer
from langchain.llms import Tongyi
import os

# 设置api_key
os.environ["DASHSCOPE_API_KEY"] = "sk-key"
llm = Tongyi()

#链接neo4j 账号密码
username = "user"
password = "password"
url = "your-url"
database = "neo4j"

graph = Neo4jGraph(
    url=url,
    username=username,
    password=password
)

# LLMGraphTransformer模块 构建图谱
llm_transformer = LLMGraphTransformer(llm=llm)
# 导入文档————参考链接中居里夫人那段文本
raw_documents = TextLoader(r"E:\task1\data\1.txt").load()
# 将文本分割成每个包含20个tokens的块，并且这些块之间没有重叠
text_splitter = CharacterTextSplitter.from_tiktoken_encoder(
    chunk_size=20, chunk_overlap=0
)
# Chunk the document
documents = text_splitter.split_documents(raw_documents)
# 打印原始文档内容
print(f"Raw Document Content: {raw_documents[0].page_content}")
# 打印分块后的文档内容
for i, doc in enumerate(documents):
    print(f"Chunk {i}: {doc.page_content}")
# 打印转换过程中的中间结果
graph_documents = llm_transformer.convert_to_graph_documents(documents)
for i, graph_doc in enumerate(graph_documents):
    print(f"Graph Document {i}:")
    print(f"Nodes: {graph_doc.nodes}")
    print(f"Relationships: {graph_doc.relationships}")

以上是运行结果，可以看到提取的节点大多都囊括了。
但是对于一个较长的语句，是我目前不需要的。

3.后续需要调整代码，限制节点的类型，以及关系的类型。（根据参考链接进行修改）

#!/usr/bin/python
# -*- coding: <utf-8> -*-
from langchain.docstore.document import Document
from langchain.text_splitter import CharacterTextSplitter
from langchain_community.document_loaders import TextLoader
from langchain_community.vectorstores import Neo4jVector
from langchain_openai import OpenAIEmbeddings, ChatOpenAI
from langchain_community.graphs import Neo4jGraph
from langchain_experimental.graph_transformers import LLMGraphTransformer
from langchain.llms import Tongyi
import os

# 设置api_key
os.environ["DASHSCOPE_API_KEY"] = "sk-key"
llm = Tongyi()

#链接neo4j 账号密码
username = "user"
password = "password"
url = "your-url"
database = "neo4j"

graph = Neo4jGraph(
    url=url,
    username=username,
    password=password
)

# 导入文档 
raw_documents = TextLoader(r"E:\task1\data\1.txt").load()
# 将文本分割成每个包含20个tokens的块，并且这些块之间没有重叠
text_splitter = CharacterTextSplitter.from_tiktoken_encoder(
    chunk_size=20, chunk_overlap=0
)
# Chunk the document
documents = text_splitter.split_documents(raw_documents)
# 打印原始文档内容
print(f"Raw Document Content: {raw_documents[0].page_content}")
# 打印分块后的文档内容
for i, doc in enumerate(documents):
    print(f"Chunk {i}: {doc.page_content}")
# 打印转换过程中的中间结果
llm_transformer_filtered = LLMGraphTransformer(
    llm=llm,
    allowed_nodes=["Person", "Country", "Organization"],
    allowed_relationships=["NATIONALITY", "LOCATED_IN", "WORKED_AT", "SPOUSE"],
)
graph_documents_filtered = llm_transformer_filtered.convert_to_graph_documents(
    documents
)
for i, graph_doc in enumerate(graph_documents_filtered):
    print(f"Graph Document {i}:")
    print(f"Nodes: {graph_doc.nodes}")
    print(f"Relationships: {graph_doc.relationships}")

以上是限制好了节点的类型，关系有哪些。

4.最后添加到你链接的Neo4j图谱中。

graph.add_graph_documents(
  graph_documents_filtered, 
  baseEntityLabel=True, 
  include_source=True
)

5.构建非结构化检索（verctor_index）

# 文本向量化 混合检索 文本和关键词检索器构建
vector_index = Neo4jVector.from_existing_graph(
    embeddings,
    search_type="hybrid",
    node_label="Document",
    text_node_properties=["text"],
    embedding_node_property="embedding"
)

6.结构化检索器（图谱）

① 提取question中的实体，后续用于图谱中去提取结构化的那些数据

from langchain.prompts import (
    PromptTemplate,
    ChatPromptTemplate,
    HumanMessagePromptTemplate,
)
from langchain.output_parsers import PydanticOutputParser
from pydantic import BaseModel, Field
from typing import List
from langchain.llms import OpenAI  # 假设使用OpenAI的LLM
import json

# Extract entities from text 定义抽取实体的类
class Entities(BaseModel):
    """Identifying information about entities."""
    names: List[str] = Field(
        ...,
        description="All the person, organization, or business entities that "
        "appear in the text",
    )

    @classmethod
    def from_json(cls, json_data: dict):
        names = []
        if 'entities' in json_data:
            for entity in json_data['entities']:
                names.append(entity['entity'])
        return cls(names=names)

entity_query = "Where did Marie Curie work?"
# Set up a parser + inject instructions into the prompt template.
parser = PydanticOutputParser(pydantic_object=Entities)

# 更新 ChatPromptTemplate 以包含新的提示词
chat_prompt = ChatPromptTemplate.from_messages(
    [
        (
            "system",
            "You are extracting organization and person entities from the text.",
        ),
        (
            "human",
            "Use the given format to extract information from the following "
            "input: {question}. Please ensure the output is in valid JSON format.",
        ),
    ]
)

# 创建提示模板
prompt_template = PromptTemplate(
    template="Answer the user query.\n{format_instructions}\n{query}\n",
    input_variables=["query"],
    partial_variables={"format_instructions": parser.get_format_instructions()},
)

# 格式化输入
_input = chat_prompt.format_prompt(question=entity_query)

# 获取 LLM 输出
output = llm(_input.to_string())

# 解析输出
parsed_output = json.loads(output)

# 使用自定义的 from_json 方法解析输出
entities = Entities.from_json(parsed_output)

# 获取 names 列表
names = entities.names
print(names)

② 构建结构化的检索器

def generate_full_text_query(input: str) -> str:
    """
    Generate a full-text search query for a given input string.

    This function constructs a query string suitable for a full-text search.
    It processes the input string by splitting it into words and appending a
    similarity threshold (~2 changed characters) to each word, then combines
    them using the AND operator. Useful for mapping entities from user questions
    to database values, and allows for some misspelings.
    """
    full_text_query = ""
    words = [el for el in remove_lucene_chars(input).split() if el]
    for word in words[:-1]:
        full_text_query += f" {word}~2 AND"
    full_text_query += f" {words[-1]}~2"
    return full_text_query.strip()

# Fulltext index query
def structured_retriever(names: List) -> str:
    """
    Collects the neighborhood of entities mentioned
    in the question
    """
    result = ""
    # # 前面提取到的实体
    # entities = entity_chain.invoke({"question": question})
    # entities = names
    # for entity in entities.names:
    for entity in names:
        response = graph.query(
            """CALL db.index.fulltext.queryNodes('entity', $query, {limit:2})
            YIELD node,score
            CALL {
              WITH node
              MATCH (node)-[r:!MENTIONS]->(neighbor)
              RETURN node.id + ' - ' + type(r) + ' -> ' + neighbor.id AS output
              UNION ALL
              WITH node
              MATCH (node)<-[r:!MENTIONS]-(neighbor)
              RETURN neighbor.id + ' - ' + type(r) + ' -> ' +  node.id AS output
            }
            RETURN output LIMIT 50
            """,
            {"query": generate_full_text_query(entity)},
        )
        result += "\n".join([el['output'] for el in response])
    return result

7.将非结构化检索器和结构化检索结合在一起

def retriever(question: str,names:List):
    print(f"Search query: {question}")
    structured_data = structured_retriever(names)
    unstructured_data = [el.page_content for el in vector_index.similarity_search(question)]
    final_data = f"""Structured data:
{structured_data}
Unstructured data:
{"#Document ". join(unstructured_data)}
    """
    return final_data

8.最后构建RAG链

# Condense a chat history and follow-up question into a standalone question
_template = """Given the following conversation and a follow up question, rephrase the follow up question to be a standalone question,
in its original language.
Chat History:
{chat_history}
Follow Up Input: {question}
Standalone question:"""  # noqa: E501
CONDENSE_QUESTION_PROMPT = PromptTemplate.from_template(_template)

def _format_chat_history(chat_history: List[Tuple[str, str]]) -> List:
    buffer = []
    for human, ai in chat_history:
        buffer.append(HumanMessage(content=human))
        buffer.append(AIMessage(content=ai))
    return buffer

_search_query = RunnableBranch(
    # If input includes chat_history, we condense it with the follow-up question
    (
        RunnableLambda(lambda x: bool(x.get("chat_history"))).with_config(
            run_name="HasChatHistoryCheck"
        ),  # Condense follow-up question and chat into a standalone_question
        RunnablePassthrough.assign(
            chat_history=lambda x: _format_chat_history(x["chat_history"])
        )
        | CONDENSE_QUESTION_PROMPT
        | llm
        | StrOutputParser(),
    ),
    # Else, we have no chat history, so just pass through the question
    RunnableLambda(lambda x : x["question"]),
)


template = """Answer the question based only on the following context:
{context}

Question: {question}
Use natural language and be concise.
Answer:"""
prompt = ChatPromptTemplate.from_template(template)

chain = (
    RunnableParallel(
        {
            "context": _search_query | retriever,
            "question": RunnablePassthrough(),
        }
    )
    | prompt
    | llm
    | StrOutputParser()
)

提问

chain.invoke({"question": "Where did Marie Curie work?"})

最后实现了整个graphRAG的链路。

参考文档：https://blog.langchain.dev/enhancing-rag-based-applications-accuracy-by-constructing-and-leveraging-knowledge-graphs/
https://blog.csdn.net/KQe397773106/article/details/138051927
源代码参考复现：https://github.com/tomasonjo/blogs/blob/master/llm/enhancing_rag_with_graph.ipynb?ref=blog.langchain.dev