Issue with Querying Neo4j Graph Database using LlamaIndex - Getting Empty Response

[NLATH]

Hi everyone,
I have created a knowledge graph in Neo4j using the neo4j framework with the following setup:
kg_builder = SimpleKGPipeline(
llm=llm,
driver=driver,
text_splitter=FixedSizeSplitter(chunk_size=500, chunk_overlap=100),
embedder=embedder,
entities=node_labels,
relations=rel_types,
prompt_template=prompt_template,
from_pdf=False # Make sure this is set to False
)
The graph is correctly built, and I can successfully query it using Neo4j directly. However, I am now trying to query this graph using the LlamaIndex framework and encountering an issue where the response is empty.
Here’s my code:
from llama_index.graph_stores.neo4j import Neo4jPropertyGraphStore
from llama_index.indices.property_graph import PropertyGraphIndex
from llama_index import OpenAIEmbedding
from llama_index.query_engine import QueryEngine

graph_store = Neo4jPropertyGraphStore(
username=NEO4J_USERNAME,
password=NEO4J_PASSWORD,
url=NEO4J_URI,
database="neo4j",
)

index = PropertyGraphIndex.from_existing(
property_graph_store=graph_store,
llm=OpenAI(model="gpt-4", temperature=0.3),
embed_model=OpenAIEmbedding(),
)

query_engine = index.as_query_engine(include_text=True)

response = await query_engine.aquery("question")

print(response)

Problem:
When I run this query, I get an empty response.
However, when I query the Neo4j graph directly using the neo4j driver, I get the expected results.
Troubleshooting so far:
Verified that the graph data is correctly populated in Neo4j.
Checked the connection details (username, password, URI) — all correct.
Tried with and without include_text=True, but still no results.
Questions:
Is there something I am missing while querying the Neo4j graph using LlamaIndex?
Do I need to re-index the graph or perform any additional steps before querying?
Could there be an issue with how the property graph is being indexed?
Any guidance would be much appreciated!