如何為圖形資料庫新增語意層

您可以使用資料庫查詢從圖形資料庫（如 Neo4j）檢索資訊。一種選擇是使用 LLM 生成 Cypher 語句。雖然這種選擇提供了極佳的彈性，但該解決方案可能很脆弱，並且無法持續產生精確的 Cypher 語句。我們可以實作 Cypher 範本作為語意層中的工具，供 LLM 代理程式互動，而不是生成 Cypher 語句。

設定

首先，取得所需的套件並設定環境變數

%pip install --upgrade --quiet  langchain langchain-neo4j langchain-openai

在本指南中，我們預設使用 OpenAI 模型，但您可以將其替換為您選擇的模型提供商。

import getpass
import os

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

# Uncomment the below to use LangSmith. Not required.
# os.environ["LANGSMITH_API_KEY"] = getpass.getpass()
# os.environ["LANGSMITH_TRACING"] = "true"

 ········

接下來，我們需要定義 Neo4j 憑證。請按照這些安裝步驟設定 Neo4j 資料庫。

os.environ["NEO4J_URI"] = "bolt://127.0.0.1:7687"
os.environ["NEO4J_USERNAME"] = "neo4j"
os.environ["NEO4J_PASSWORD"] = "password"

以下範例將建立與 Neo4j 資料庫的連線，並使用關於電影及其演員的範例資料填充它。

from langchain_neo4j import Neo4jGraph

graph = Neo4jGraph(refresh_schema=False)

# Import movie information

movies_query = """
LOAD CSV WITH HEADERS FROM 
'https://raw.githubusercontent.com/tomasonjo/blog-datasets/main/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)

API 參考：Neo4jGraph

[]

使用 Cypher 範本的自訂工具

語意層由暴露給 LLM 的各種工具組成，LLM 可以使用這些工具與知識圖譜互動。它們的複雜程度可能各不相同。您可以將語意層中的每個工具視為一個函數。

我們將實作的函數是檢索關於電影或其演員陣容的資訊。

description_query = """
MATCH (m:Movie|Person)
WHERE m.title CONTAINS $candidate OR m.name CONTAINS $candidate
MATCH (m)-[r:ACTED_IN|IN_GENRE]-(t)
WITH m, type(r) as type, collect(coalesce(t.name, t.title)) as names
WITH m, type+": "+reduce(s="", n IN names | s + n + ", ") as types
WITH m, collect(types) as contexts
WITH m, "type:" + labels(m)[0] + "\ntitle: "+ coalesce(m.title, m.name) 
       + "\nyear: "+coalesce(m.released,"") +"\n" +
       reduce(s="", c in contexts | s + substring(c, 0, size(c)-2) +"\n") as context
RETURN context LIMIT 1
"""


def get_information(entity: str) -> str:
    try:
        data = graph.query(description_query, params={"candidate": entity})
        return data[0]["context"]
    except IndexError:
        return "No information was found"

您可以觀察到我們已定義用於檢索資訊的 Cypher 語句。因此，我們可以避免生成 Cypher 語句，而僅使用 LLM 代理程式來填充輸入參數。為了向 LLM 代理程式提供關於何時使用工具及其輸入參數的額外資訊，我們將該函數包裝為工具。

from typing import Optional, Type

from langchain_core.tools import BaseTool
from pydantic import BaseModel, Field


class InformationInput(BaseModel):
    entity: str = Field(description="movie or a person mentioned in the question")


class InformationTool(BaseTool):
    name: str = "Information"
    description: str = (
        "useful for when you need to answer questions about various actors or movies"
    )
    args_schema: Type[BaseModel] = InformationInput

    def _run(
        self,
        entity: str,
    ) -> str:
        """Use the tool."""
        return get_information(entity)

    async def _arun(
        self,
        entity: str,
    ) -> str:
        """Use the tool asynchronously."""
        return get_information(entity)

API 參考：BaseTool

LangGraph 代理程式

我們將使用 LangGraph 實作一個簡單的 ReAct 代理程式。

代理程式由 LLM 和工具步驟組成。當我們與代理程式互動時，我們將首先呼叫 LLM 以決定是否應使用工具。然後我們將執行一個迴圈

如果代理程式表示要採取動作（即呼叫工具），我們將執行工具並將結果傳遞回代理程式。如果代理程式未要求執行工具，我們將完成（回應用戶）。

程式碼實作非常簡單。首先，我們將工具繫結到 LLM 並定義助理步驟。

from langchain_core.messages import HumanMessage, SystemMessage
from langchain_openai import ChatOpenAI
from langgraph.graph import MessagesState

llm = ChatOpenAI(model="gpt-4o")

tools = [InformationTool()]
llm_with_tools = llm.bind_tools(tools)

# System message
sys_msg = SystemMessage(
    content="You are a helpful assistant tasked with finding and explaining relevant information about movies."
)


# Node
def assistant(state: MessagesState):
    return {"messages": [llm_with_tools.invoke([sys_msg] + state["messages"])]}

API 參考：HumanMessage | SystemMessage | ChatOpenAI

接下來，我們定義 LangGraph 流程。

from IPython.display import Image, display
from langgraph.graph import END, START, StateGraph
from langgraph.prebuilt import ToolNode, tools_condition

# Graph
builder = StateGraph(MessagesState)

# Define nodes: these do the work
builder.add_node("assistant", assistant)
builder.add_node("tools", ToolNode(tools))

# Define edges: these determine how the control flow moves
builder.add_edge(START, "assistant")
builder.add_conditional_edges(
    "assistant",
    # If the latest message (result) from assistant is a tool call -> tools_condition routes to tools
    # If the latest message (result) from assistant is a not a tool call -> tools_condition routes to END
    tools_condition,
)
builder.add_edge("tools", "assistant")
react_graph = builder.compile()

# Show
display(Image(react_graph.get_graph(xray=True).draw_mermaid_png()))

API 參考：StateGraph | ToolNode | tools_condition

現在，讓我們用一個範例問題測試工作流程。

input_messages = [HumanMessage(content="Who played in the Casino?")]
messages = react_graph.invoke({"messages": input_messages})
for m in messages["messages"]:
    m.pretty_print()

================================ Human Message =================================

Who played in the Casino?
================================== Ai Message ==================================
Tool Calls:
  Information (call_j4usgFStGtBM16fuguRaeoGc)
 Call ID: call_j4usgFStGtBM16fuguRaeoGc
  Args:
    entity: Casino
================================= Tool Message =================================
Name: Information

type:Movie
title: Casino
year: 1995-11-22
ACTED_IN: Robert De Niro, Joe Pesci, Sharon Stone, James Woods
IN_GENRE: Drama, Crime

================================== Ai Message ==================================

The movie "Casino," released in 1995, features the following actors:

- Robert De Niro
- Joe Pesci
- Sharon Stone
- James Woods

The film is in the Drama and Crime genres.