Chatting with knowledge graph

Neo4j Graph Platform
1

Hi everyone, i just got started on neo4j and am interested in building a chat app to chat with my knowledge graph

i used the python code below but its performance is nowhere near as good as the existing chatbot on the graph builder UI, how would go about building one as good as the one in the UI (using Graph+Vector+fulltext)

image
image585×937 79.5 KB

graph = Neo4jGraph(
    url=neo4j_uri,
    username=neo4j_username,
    password=neo4j_password
)

llm = ChatOpenAI(model="gpt-3.5-turbo")
chain = GraphCypherQAChain.from_llm(
    llm=llm,
    graph=graph,
    verbose=True,
    allow_dangerous_requests=True
)

app = FastAPI()

class QuestionRequest(BaseModel):
    question: str

@app.post("/ask")
def ask_question(request: QuestionRequest):
    try:
        response = chain.run(request.question)
        return {"answer": response}
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/", response_class=HTMLResponse)
def root():
    with open("frontend.html", "r", encoding="utf-8") as f:
        return f.read()
2

Have you had a look at the LLM Graph Builder code? GitHub - neo4j-labs/llm-graph-builder: Neo4j graph construction from unstructured data using LLMs