Question : how to insert CypherQueryCorrector for chaining

[vishnu1729]

Hi ,

In one of my projects, I am intending to convert user's Natural Language questions into Cypher queries. I found this example to be really useful and mostly following the same structure - https://github.com/neo4j-labs/text2cypher/blob/main/evaluations/evaluating_cypher_jaccard.ipynb.

My LLM model is deployed in Azure Open AI and this is how I have set things up

token_provider = get_bearer_token_provider(DefaultAzureCredential(exclude_interactive_browser_credential=False), "https://cognitiveservices.azure.com/.default")
 
llm = AzureChatOpenAI(
    api_version='<api_version>',
    azure_endpoint="<azure_deployment_endpoint>",
    azure_ad_token_provider=token_provider,
    verbose = True
)

# Generate Cypher statement based on natural language input
cypher_template = """Based on the Neo4j graph schema below,
write a Cypher query that would answer the user's question.
Return only Cypher statement, no backticks, nothing else.
{schema}

Question: {question}
Cypher query:"""  # noqa: E501

cypher_prompt = ChatPromptTemplate.from_messages(
    [
        (
            "system",
            "Given an input question, convert it to a Cypher query. No pre-amble.",
        ),
        ("human", cypher_template),
    ]
)

# https://python.langchain.com/docs/how_to/sequence/
cypher_chain = (
    RunnablePassthrough.assign(
        schema=lambda _: graph.get_schema,
    )
    | cypher_prompt
    | llm.bind(stop=["\nCypherResult:"])
    | StrOutputParser()
)

As you can see I am mostly following the notebook's structure for my work as i am new to Neo4j and using LLMs.
In the linked notebook, i see that CypherQueryCorrector is imported but not used. I need this now as some of my prompts are generating Cyphers with wrong syntax.
Can you advise how this can be inserted in the chain structure above?