Duplicate Master Records issue

Neo4j Graph Platform
1

My current schema is as below where a "Customer_ERP_Auto" node (in brown) gets connected to other related "Customer_ERP_Auto" nodes via multiple relationships.

sethus_0-1674192253832.png

My Objective here is to create a new “Customer_Master” node for every such group / cluster of nodes connected using the "FUZZY_SIMILARITY" relationship. The newly created "Customer_Master" node has to link with all the nodes in the group using the “HAS_MASTER” relationship as shown below.

sethus_1-1674192393464.png

I tried creating the below Cypher query for this purpose.

call apoc.periodic.iterate(

"MATCH(a:Customer_ERP_Auto)-[r:FUZZY_SIMILARITY]->(b:Customer_ERP_Auto)

WHERE NOT exists((a)-[:HAS_MASTER]->())

return a,b",

"WITH a, collect(b.customer_id) as customer_ids

WITH a, [cust_id in customer_ids where NOT EXISTS((:Customer_ERP_Auto {customer_id:cust_id})-[:HAS_MASTER]->())] as f_customer_ids

UNWIND f_customer_ids as f_cust_id

MERGE(m:Customer_Master {master_id:a.customer_id})

MERGE(c:Customer_ERP_Auto {customer_id:f_cust_id})

MERGE(c)-[:HAS_MASTER]->(m)",

{parallel:False});

But the above query ends up creating duplicate Customer_Master records as shown below.

sethus_2-1674192426710.png

I would need your help in fixing the above query such that there is only ONE “Customer_Master” for every Customer_ERP_Auto node in the DB.

Looking forward to your valuable inputs.

Thanks in advance.

2

Hi @glilienfield,
Thanks for your quick response. Conceptually, I want a Customer_ERP_Auto node to get connected to only one common "main" node. I use the "FUZZY_SIMILARITY" relationship to identify related nodes to a give node 'a' as I want to associate the node 'a' and all its associated nodes using "FUZZY_SIMILARITY" relationship to get connected to the same "Customer_Master" node using the "HAS_MASTER" relationship.

Appreciate if you could help me in correcting the above cypher query to fulfill this requirement.

3

The first line in your first query is going to result in multiple rows with different values for 'a'. Assuming there are no has_master relationships yet, you will then get many rows with multiple values of 'a' from the 'data driven' query of the iterate procedure.

In the processing query, all the customer_ids values for each 'a' will pass through the list comprehension operation because no has_master relationship has been created yet. The result will be one main node for each value of 'a' and a relationship from each new main node to each customer_id nodes corresponding to its 'a' node.

I am not sure of the requirement. it seems in your conceptional figure that you just are connecting each node to a common 'main' node. How does the FUZZY_SIMILARITY relationship influent what gets connected to a main node?