Failed to invoke procedure `gds.graph.project`

Neo4j Graph Platform
1

I try to use Neo4j to find clients with similar purchases. My test DB is:

CREATE (nMike:Client {name:'Mike'})
CREATE (nAlice:Client {name:'Alice'})
CREATE (nMark:Client {name:'Mark'})
CREATE (nCharles:Client {name:'Charles'})


CREATE (nMilk:Product {product_name:'Milk'})
CREATE (nBread:Product {product_name:'Bread'})
CREATE (nBeer:Product {product_name:'Beer'})
CREATE (nSugar:Product {product_name:'Sugar'})
CREATE (nFlour:Product {product_name:'Flour'})

CREATE (nMike)-[:BOUGHT]->(nMilk)
CREATE (nMike)-[:BOUGHT]->(nBread)
CREATE (nMike)-[:BOUGHT]->(nBeer)
CREATE (nMike)-[:BOUGHT]->(nSugar)

CREATE (nAlice)-[:BOUGHT]->(nMilk)
CREATE (nAlice)-[:BOUGHT]->(nBread)

CREATE (nMark)-[:BOUGHT]->(nBread)
CREATE (nMark)-[:BOUGHT]->(nBeer)

CREATE (nCharles)-[:BOUGHT]->(nMilk)
CREATE (nCharles)-[:BOUGHT]->(nFlour);

Now I try to create graph. If I get it correctly, 'product_name' property matters - I want to find clients with similar product purchase.
So I try to execute code:

CALL gds.graph.project(
  'graph',                                
  {                                                     
    Client: {properties: 'name'},                        
    Product: {properties: 'product_name'}    
  },
  {BOUGHT: {orientation: 'UNDIRECTED'}}                      
)

But it fails: Failed to invoke procedure gds.graph.project: Caused by: java.lang.UnsupportedOperationException: Loading of values of type String is currently not supported
What is wrong here? Looks exactly how it should be: https://neo4j.com/docs/graph-data-science/current/management-ops/graph-creation/graph-project/
In fact I do not need 'name' property for Client nodes but getting error if empty.

Some suggestions please?

2

It looks like GDS only supports projecting numeric properties.

Screen Shot 2023-12-03 at 1.37.15 PM
Screen Shot 2023-12-03 at 1.37.15 PM1512×872 67.9 KB

You could encode them by mapping each product name to an integer. Don't project the name property. You can get the client names from the results by using the gds.util.asNode() method to get each node from the id returned, then the name property from the node.

3

Thank you.
So I added property 'product_id' as integer. But have no idea how to create this graph.

CALL gds.graph.project(
  'graph',
  'Client',                             
  {                                                                            
    Product: {properties: 'product_id'}    
  },
  {BOUGHT: {orientation: 'UNDIRECTED'}}                      
)

Wrong

CALL gds.graph.project(
  'graph',                                
  {                                                     
    Client: ,                        
    Product: {properties: 'product_id'}    
  },
  {BOUGHT: {orientation: 'UNDIRECTED'}}                      
)

Wrong
How to include 'Client' nodes (without properties) in this graph?

Anyway gds.louvain.stream algorithm should give me communities of clients with similar purchases (similar 'product_id's), right?

4

After next doezens of tries I have graph project syntax:

CALL gds.graph.project(
  'graph',
  ['Client', 'Product'],
  ['BOUGHT'],
  {nodeProperties: 'product_id'}
)

And Louvian algorithim (I do not get why it returns 'null'):

CALL gds.louvain.stream('graph')
 YIELD nodeId, communityId, intermediateCommunityIds 
 WHERE gds.util.asNode(nodeId).name is not null
 RETURN gds.util.asNode(nodeId).name AS name, communityId 
 ORDER BY name ASC

Result looks quite resonible but I do not know if this is what I'm looking for