Hi, I've got a problem with llamaindex,

This is one of the place. Shoot your query

I've got an existing neo4j graph with some nodes and relationships between them.

The nodes "Chunks" have the following properties:

• text: wich contains a text chunk
• embedding: which contains a vector representation (1536) dimension of text

Now I want to attach LlamaIndex with this existing graph and the existing index build with neo4j.

So i can do the following:

neo4j_vector = Neo4jVectorStore(
embedding_dimension=1536,
username=os.getenv("NEO4J_USERNAME"),
password=os.getenv("NEO4J_PASSWORD"),
url=os.getenv("NEO4J_URL"),
index_name=VECTOR_INDEX_NAME,
text_node_property=VECTOR_SOURCE_PROPERTY,
embedding_node_property='embedding',
node_label=VECTOR_NODE_LABEL,
)

retriever = VectorStoreIndex.from_vector_store(neo4j_vector).as_retriever()

retriever.retrieve("my query").

But I've got the following error:
ValueError("Node content not found in metadata dict.")

I think that llama is expecting a metadata dictionary in the nodes (because i didn't creare the embeddings and the nodes with llama).

How can i resolve this?

I think best way to resolve this would be to create the embeddings again as the structure for storing the content might be diff for other application that you may have used

That is my idea, but how can i do that?

I would like to use llama to iterate over my neo4j nodes and then add the structure and embedding that llama wants

How did you created these nodes

via chyper

using the neo4j python driver

then, i create the index with this query:
CREATE VECTOR INDEX chunks_embedding IF NOT EXISTS
FOR (c:Chunk) ON (c.embedding)
OPTIONS { indexConfig: {
vector.dimensions: 1536,
vector.similarity_function: 'cosine'
}}

MATCH (c:Chunk) WHERE c.text IS NOT NULL
WITH c, genai.vector.encode(
c.text,
"OpenAI",
{
token: MY_TOKEN,
endpoint: "https://api.openai.com/v1/embeddings"
}) AS vector
CALL db.create.setNodeVectorProperty(c, "embedding", vector)

No idea about this tbh, I would suggest you create the nodes following this doc: https://docs.llamaindex.ai/en/stable/examples/vector_stores/Neo4jVectorDemo/?h=neo4j

That way it will be compataible with everything you do with llamaindex

Ok got it, but then i need to create some relationships between the nodes generated by llama and other nodes inside the graph, is it possible right?