ym88659208ym87991671
Neo4j DB QA chain | Документация для разработчиков

Neo4j DB QA chain

Обновлено 6 марта 2024

This notebook shows how to use LLMs to provide a natural language interface to a graph database you can query with the Cypher query language. You will need to have a running Neo4j instance. One option is to create a free Neo4j database instance in their Aura cloud service. You can also run the database locally using the Neo4j Desktop application, or running a docker container. You can run a local docker container by running the executing the following script:

docker run \
--name neo4j \
-p 7474:7474 -p 7687:7687 \
-d \
-e NEO4J_AUTH=neo4j/pleaseletmein \
-e NEO4J_PLUGINS=\[\"apoc\"\] \
neo4j:latest

If you are using the docker container, you need to wait a couple of second for the database to start.

from langchain.chains import GraphCypherQAChain
from langchain_community.graphs import Neo4jGraph
from langchain_openai import ChatOpenAI
graph = Neo4jGraph(
url="bolt://localhost:7687", username="neo4j", password="pleaseletmein"
)

Seeding the database

Assuming your database is empty, you can populate it using Cypher query language. The following Cypher statement is idempotent, which means the database information will be the same if you run it one or multiple times.

graph.query(
"""
MERGE (m:Movie {name:"Top Gun"})
WITH m
UNWIND ["Tom Cruise", "Val Kilmer", "Anthony Edwards", "Meg Ryan"] AS actor
MERGE (a:Actor {name:actor})
MERGE (a)-[:ACTED_IN]->(m)
"""
)
    []

Refresh graph schema information

If the schema of database changes, you can refresh the schema information needed to generate Cypher statements.

graph.refresh_schema()
print(graph.schema)
    Node properties are the following:
Movie {name: STRING},Actor {name: STRING}
Relationship properties are the following:

The relationships are the following:
(:Actor)-[:ACTED_IN]->(:Movie)

Querying the graph

We can now use the graph cypher QA chain to ask question of the graph

chain = GraphCypherQAChain.from_llm(
ChatOpenAI(temperature=0), graph=graph, verbose=True
)
chain.run("Who played in Top Gun?")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})
RETURN a.name
Full Context:
[{'a.name': 'Tom Cruise'}, {'a.name': 'Val Kilmer'}, {'a.name': 'Anthony Edwards'}, {'a.name': 'Meg Ryan'}]

> Finished chain.
    'Tom Cruise, Val Kilmer, Anthony Edwards, and Meg Ryan played in Top Gun.'

Limit the number of results

You can limit the number of results from the Cypher QA Chain using the top_k parameter. The default is 10.

chain = GraphCypherQAChain.from_llm(
ChatOpenAI(temperature=0), graph=graph, verbose=True, top_k=2
)
chain.run("Who played in Top Gun?")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})
RETURN a.name
Full Context:
[{'a.name': 'Tom Cruise'}, {'a.name': 'Val Kilmer'}]

> Finished chain.
    'Tom Cruise and Val Kilmer played in Top Gun.'

Return intermediate results

You can return intermediate steps from the Cypher QA Chain using the return_intermediate_steps parameter

chain = GraphCypherQAChain.from_llm(
ChatOpenAI(temperature=0), graph=graph, verbose=True, return_intermediate_steps=True
)
result = chain("Who played in Top Gun?")
print(f"Intermediate steps: {result['intermediate_steps']}")
print(f"Final answer: {result['result']}")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})
RETURN a.name
Full Context:
[{'a.name': 'Tom Cruise'}, {'a.name': 'Val Kilmer'}, {'a.name': 'Anthony Edwards'}, {'a.name': 'Meg Ryan'}]

> Finished chain.
Intermediate steps: [{'query': "MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})\nRETURN a.name"}, {'context': [{'a.name': 'Tom Cruise'}, {'a.name': 'Val Kilmer'}, {'a.name': 'Anthony Edwards'}, {'a.name': 'Meg Ryan'}]}]
Final answer: Tom Cruise, Val Kilmer, Anthony Edwards, and Meg Ryan played in Top Gun.

Return direct results

You can return direct results from the Cypher QA Chain using the return_direct parameter

chain = GraphCypherQAChain.from_llm(
ChatOpenAI(temperature=0), graph=graph, verbose=True, return_direct=True
)
chain.run("Who played in Top Gun?")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})
RETURN a.name

> Finished chain.
    [{'a.name': 'Tom Cruise'},
{'a.name': 'Val Kilmer'},
{'a.name': 'Anthony Edwards'},
{'a.name': 'Meg Ryan'}]

Add examples in the Cypher generation prompt

You can define the Cypher statement you want the LLM to generate for particular questions

from langchain.prompts.prompt import PromptTemplate

CYPHER_GENERATION_TEMPLATE = """Task:Generate Cypher statement to query a graph database.
Instructions:
Use only the provided relationship types and properties in the schema.
Do not use any other relationship types or properties that are not provided.
Schema:
{schema}
Note: Do not include any explanations or apologies in your responses.
Do not respond to any questions that might ask anything else than for you to construct a Cypher statement.
Do not include any text except the generated Cypher statement.
Examples: Here are a few examples of generated Cypher statements for particular questions:
# How many people played in Top Gun?
MATCH (m:Movie {{title:"Top Gun"}})<-[:ACTED_IN]-()
RETURN count(*) AS numberOfActors

The question is:
{question}"""

CYPHER_GENERATION_PROMPT = PromptTemplate(
input_variables=["schema", "question"], template=CYPHER_GENERATION_TEMPLATE
)

chain = GraphCypherQAChain.from_llm(
ChatOpenAI(temperature=0),
graph=graph,
verbose=True,
cypher_prompt=CYPHER_GENERATION_PROMPT,
)
chain.run("How many people played in Top Gun?")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (m:Movie {name:"Top Gun"})<-[:ACTED_IN]-(:Actor)
RETURN count(*) AS numberOfActors
Full Context:
[{'numberOfActors': 4}]

> Finished chain.
    'Four people played in Top Gun.'

Use separate LLMs for Cypher and answer generation

You can use the cypher_llm and qa_llm parameters to define different llms

chain = GraphCypherQAChain.from_llm(
graph=graph,
cypher_llm=ChatOpenAI(temperature=0, model="gpt-3.5-turbo"),
qa_llm=ChatOpenAI(temperature=0, model="gpt-3.5-turbo-16k"),
verbose=True,
)
chain.run("Who played in Top Gun?")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})
RETURN a.name
Full Context:
[{'a.name': 'Tom Cruise'}, {'a.name': 'Val Kilmer'}, {'a.name': 'Anthony Edwards'}, {'a.name': 'Meg Ryan'}]

> Finished chain.
    'Tom Cruise, Val Kilmer, Anthony Edwards, and Meg Ryan played in Top Gun.'

Ignore specified node and relationship types

Обновлено 6 марта 2024

You can use include_types or exclude_types to ignore parts of the graph schema when generating Cypher statements.

chain = GraphCypherQAChain.from_llm(
graph=graph,
cypher_llm=ChatOpenAI(temperature=0, model="gpt-3.5-turbo"),
qa_llm=ChatOpenAI(temperature=0, model="gpt-3.5-turbo-16k"),
verbose=True,
exclude_types=["Movie"],
)
# Inspect graph schema
print(chain.graph_schema)
    Node properties are the following:
Actor {name: STRING}
Relationship properties are the following:

The relationships are the following:

Validate generated Cypher statements

Обновлено 6 марта 2024

You can use the validate_cypher parameter to validate and correct relationship directions in generated Cypher statements

chain = GraphCypherQAChain.from_llm(
llm=ChatOpenAI(temperature=0, model="gpt-3.5-turbo"),
graph=graph,
verbose=True,
validate_cypher=True,
)
chain.run("Who played in Top Gun?")
    

> Entering new GraphCypherQAChain chain...
Generated Cypher:
MATCH (a:Actor)-[:ACTED_IN]->(m:Movie {name: 'Top Gun'})
RETURN a.name
Full Context:
[{'a.name': 'Tom Cruise'}, {'a.name': 'Val Kilmer'}, {'a.name': 'Anthony Edwards'}, {'a.name': 'Meg Ryan'}]

> Finished chain.
    'Tom Cruise, Val Kilmer, Anthony Edwards, and Meg Ryan played in Top Gun.'
ПАО Сбербанк использует cookie для персонализации сервисов и удобства пользователей.
Вы можете запретить сохранение cookie в настройках своего браузера.