Neo4j extension

Usage

The Neo4j extension allows you to migrate your Neo4j data into a Kuzu database. The functionality is not available by default, so you first need to install and load the Neo4j extension by running the following commands:

INSTALL neo4j;
LOAD neo4j;

Note that migrating multi-labelled nodes from Neo4j to Kuzu is currently not supported since Kuzu doesn’t support multi-labelled nodes.

Example dataset

We’ll use the following queries to create an example graph in Neo4j to demonstrate how this extension can be used. First, set up your Neo4j Desktop locally or a Neo4j database connection, and run the following Cypher queries:

// Create person nodes
CREATE (u:User {name: 'Adam', age: 30});
CREATE (u:User {name: 'Karissa', age: 40});
CREATE (u:User {name: 'Zhang', age: 50});
CREATE (u:User {name: 'Noura', age: 25});
// Create city nodes
CREATE (c:City {name: 'Waterloo', population: 150000});
CREATE (c:City {name: 'Kitchener', population: 200000});
CREATE (c:City {name: 'Guelph', population: 75000});
// Create person-city relationships
MATCH (u:User {name: 'Adam'}), (c:City {name: 'Waterloo'}) CREATE (u)-[:LivesIn]->(c);
MATCH (u:User {name: 'Karissa'}), (c:City {name: 'Waterloo'}) CREATE (u)-[:LivesIn]->(c);
MATCH (u:User {name: 'Zhang'}), (c:City {name: 'Kitchener'}) CREATE (u)-[:LivesIn]->(c);
MATCH (u:User {name: 'Noura'}), (c:City {name: 'Guelph'}) CREATE (u)-[:LivesIn]->(c);
// Create person-person relationships
MATCH (u:User {name: 'Adam'}), (u1:User {name: 'Karissa'}) CREATE (u)-[:Follows {since: 2020}]->(u1);
MATCH (u:User {name: 'Adam'}), (u1:User {name: 'Zhang'}) CREATE (u)-[:Follows {since: 2020}]->(u1);
MATCH (u:User {name: 'Karissa'}), (u1:User {name: 'Zhang'}) CREATE (u)-[:Follows {since: 2021}]->(u1);
MATCH (u:User {name: 'Zhang'}), (u1:User {name: 'Noura'}) CREATE (u)-[:Follows {since: 2022}]->(u1);

The queries above create a set of nodes of label User and City, and a set of relationships of label LivesIn and Follows.

Set up the Neo4j APOC extension

The Neo4j migration functionality relies on the APOC extension to be installed in Neo4j desktop or server. You can find details on how to install APOC in Neo4j here.

Also add the following config lines to the apoc.conf file:

apoc.export.file.enabled=true
apoc.import.file.use_neo4j_config=false

This doc provides instructions on Neo4j’s default conf file locations. The apoc.conf file may not exist in the conf directory, under which case, you’d have to create a new one under the conf directory.

If you’re running Neo4j in Docker, you can use -e NEO4J_apoc_export_file_enabled=true -e NEO4J_apoc_import_file_use__neo4j__config=false to configure APOC. Additionally, you’ll need to mount /tmp as -v /tmp:/tmp to allow access to the exported data from Neo4j.

In the following section, we show how to migrate the labels from Neo4j to tables in Kuzu.

Data Migration

We provide the NEO4J_MIGRATE function to migrate data from Neo4j to Kuzu. The function takes in a few parameters specified as follows:

CALL NEO4J_MIGRATE(
    'URL',
    'USER_NAME',
    'PASSWORD',
    ['NODE_LABEL1', 'NODE_LABEL2', 'NODE_LABEL3', ...],
    ['REL_LABEL1', 'REL_LABEL2', 'REL_LABEL3', ...]
)

URL: A http://hostname:port connection url to a Neo4j server.
USER_NAME: Neo4j username. Leave blank if not needed.
PASSWORD: Neo4j password. Leave blank if not needed.
NODE_LABEL: The node labels to migrate to kuzu. (Note: multi-label nodes are not yet supported)
REL_LABEL: The rel labels to migrate to kuzu.

In our example above, we can specify User and City for node tables, and LivesIn and Follows fore relationship tables:

CALL NEO4J_MIGRATE(
    "http://localhost:7474",
    "neo4j",
    "neo4jpassword",
    ["User", "City"],
    ["Follows", "LivesIn"]
);

Once finished, you can use SHOW_TABLES() to verify that the tables have been imported into Kuzu:

CALL SHOW_TABLES() RETURN *

This should return the node and relationship tables created in Kuzu corresponding to the migrated labels from Neo4j.

┌────────┬─────────┬────────┬───────────────┬─────────┐
│ id     │ name    │ type   │ database name │ comment │
│ UINT64 │ STRING  │ STRING │ STRING        │ STRING  │
├────────┼─────────┼────────┼───────────────┼─────────┤
│ 0      │ User    │ NODE   │ local(kuzu)   │         │
│ 2      │ Follows │ REL    │ local(kuzu)   │         │
│ 1      │ City    │ NODE   │ local(kuzu)   │         │
│ 3      │ LivesIn │ REL    │ local(kuzu)   │         │
└────────┴─────────┴────────┴───────────────┴─────────┘

That’s it! The data from Neo4j should now be available in Kuzu and you can start querying them. For example, let’s find the user who was most recently followed by another user.

MATCH (u:User)-[f:Follows]->(u1:User)
RETURN u.name ORDER BY f.since DESC LIMIT 1;

Result:

┌────────┬─────────┐
│ u.name │ f.since │
│ STRING │ INT64   │
├────────┼─────────┤
│ Zhang  │ 2022    │
└────────┴─────────┘