How can I find the neighbor nodes when the nodes are not from __Entity__ type?

Neo4j Graph Platform Cypher
,
1

I am a beginner in learning about Neo4J and Cypher queries, and following this GitHub for building a Hybrid RAG using knowledge graphs. I have added a graph to my Neo4j workspace using Cypher query. The example of commands is shown below:

MERGE (nizhny_novgorod:birthPlace {id: "Nizhny Novgorod"})
MERGE (large_language_models_like_chatgpt:characteristics {id: "large language models like ChatGPT"})
MERGE (richard_zemel:person {id: "Richard Zemel"})
MERGE (mixtures_of_experts:invention {id: "mixtures of experts"})
MERGE (december_6_1947:date {id: "December 6, 1947"})
MERGE (nobel_prize_in_physics:award {id: "Nobel Prize in Physics"})

MERGE (geoffrey_hinton)-[:bornIn]->(london)
MERGE (geoffrey_hinton)-[:bornAt]->(december_6_1947)
MERGE (geoffrey_hinton)-[:worksAt]->(university_of_toronto)
MERGE (geoffrey_hinton)-[:hasNationality]->(britishcanadian)
MERGE (geoffrey_hinton)-[:hasRole]->(cognitive_psychologist_and_computer_scientist)
MERGE (geoffrey_hinton)-[:coAuthored]->(david_rumelhart)
MERGE (geoffrey_hinton)-[:coAuthored]->(ronald_j_williams)

And the problem is that the __Entity__ is empty, and it does not return any node, when I run the MATCH (n:__Entity__) RETURN n LIMIT 25; command.

I faced with this problem when trying to find neighbor nodes while retrieving the knowledge graph using the below function.

# Fulltext index query
def structured_retriever(question: str) -> str:
    """
    Or Graph Retriever
    Collects the neighborhood of entities mentioned in the question
    """
    result = ""
    entities = entity_chain.invoke({"question": question})
    print(f"Extracted entities: {entities.names}")
    for entity in entities.names:
        query = generate_full_text_query(entity)
        print(f"Generated query: {query}")
        response = graph.query(
            """CALL db.index.fulltext.queryNodes('entity', $query, {limit:2})
            YIELD node, score
            CALL {
              WITH node
              MATCH (node)-[r]->(neighbor)
              RETURN node.id + ' - ' + type(r) + ' -> ' + neighbor.id AS output
              UNION ALL
              WITH node
              MATCH (node)<-[r]-(neighbor)
              RETURN neighbor.id + ' - ' + type(r) + ' -> ' +  node.id AS output
            }
            RETURN output LIMIT 50
            """,
            {"query": query},
        )
        print(f"Query response: {response}")
        if response:
            result += "\n".join([el['output'] for el in response])
        else:
            print(f"No results found for entity: {entity}")
    return result

The above query does not return anything. For example, for the below question:

print(structured_retriever("Where was Geoffrey Hinton born"))

Output:

Extracted entities: ['Geoffrey Hinton']
Generated query: Geoffrey~2 AND Hinton~2
Query response: []
No results found for entity: Geoffrey Hinton
2

To find neighbor nodes in Neo4j when the nodes are not of the __Entity__ type, you can use Cypher queries to traverse relationships between nodes of any type. Here’s how you can do it:

Step 1: Identify the Node

First, determine how you want to identify the starting node. This could be based on a property or a specific label.

Step 2: Write the Cypher Query

You can use a query like the following to find neighbor nodes:

cypher

Copy code

MATCH (n:YourNodeLabel {yourProperty: 'yourValue'})-[:YOUR_RELATIONSHIP_TYPE]-(neighbor)
RETURN neighbor

Explanation:

  • Replace YourNodeLabel with the label of the node you are starting from.
  • Replace yourProperty and yourValue with the property and value you are using to identify the node.
  • Replace YOUR_RELATIONSHIP_TYPE with the type of relationship connecting to neighbor nodes. If you want to find all neighbors regardless of the relationship type, you can use -[]- instead.

Example Query

If you have a node labeled Person with a property name and you want to find all its neighbors, the query would look like this:

cypher

Copy code

MATCH (p:Person {name: 'Alice'})-[]-(neighbor)
RETURN neighbor

Additional Considerations

  • If you want to limit the depth of the traversal, you can specify a maximum number of relationship hops, like this:

cypher

Copy code

MATCH (n:YourNodeLabel {yourProperty: 'yourValue'})-[:YOUR_RELATIONSHIP_TYPE*1..2]-(neighbor)
RETURN neighbor
  • You can also filter neighbors based on specific criteria by adding WHERE clauses.

By using these Cypher queries, you can effectively find neighboring nodes regardless of their type.

1 Like
3

Your query looks fine. Does the full text search return results? Execute just it to see.

1 Like
4

To find the neighbor nodes when the nodes are not of the __Entity__ type in Neo4j, you can use Cypher queries that match nodes based on their properties or labels, and then traverse the relationships to find neighbors. Here's a step-by-step approach:

  1. Identify the Starting Node: First, determine which node you want to start from. This could be based on a property, label, or ID.Example: If you're looking for a node with a specific ID, you can match it like this:

cypher

Copy code

MATCH (n {id: "Geoffrey Hinton"})
  1. Match Neighboring Nodes: Use Cypher's relationship patterns to match outgoing or incoming relationships to neighboring nodes.For outgoing relationships:

cypher

Copy code

MATCH (n {id: "Geoffrey Hinton"})-[r]->(neighbor)
RETURN neighbor
LIMIT 25;

For incoming relationships:

cypher

Copy code

MATCH (n {id: "Geoffrey Hinton"})<-[r]-(neighbor)
RETURN neighbor
LIMIT 25;
  1. Use Pattern Matching: You can also match nodes based on their label or property to find neighbors of any type of node, not just __Entity__ nodes. Here's an example that matches any node with a relationship to the starting node:

cypher

Copy code

MATCH (n {id: "Geoffrey Hinton"})-[r]-(neighbor)
RETURN neighbor
LIMIT 25;

This approach ensures you can find neighbor nodes even if they are not labeled as __Entity__, as long as there is a relationship connecting them.