Recalculating Neo4J relationships based on chain of relationships

I have a long chain of relations:

:User can have multiple :Territory(ies)
:Account can belong in multiple :Territory(ies)
:Account can be assigned to a multiple :User(s)
:Account can have an :Invoice(s)
:Invoice can have an :InvoiceEntry(ies)

To sum up:

:User-[:BELONGS_TO]->:Territory<-[:EXISTS_IN]-:Account-[:ISSUED]->:Invoice-[:HAS]->:InvoiceEntry

I can simplify the relation to

:User-[:CAN_SEE]->:Invoice

Since I'm not proficient with Cypher, my questions are:

Question #1.

What would be the query to refresh the relationships between :User and :Invoice once the :Account is being removed from :Territory and leave valid relationships :CAN_SEE untouched.

Question #2.

Same as above but :User is removed from :Territory and can no longer see :Invoice.

UPDATE 16.06.23

My primary relation chain is :User -> :Territory <- :Account -> :Invoice.<br>
:User -> :Territory relation is provided by the Active Directory.<br>
:Account -> :Territory and :Account -> Invoice relations are fetched from ERP system.

In order to avoid costly calculation, because single :User can have hundreds of :Territory and one :Account can be in a hundreds of :Territory, I want to introduce a "shortcut" relationship :User-[:CAN_SEE]->:Invoice.

Example case:

U1 belongs to T1, T2, T3.<br>
A1 exists in T1, T4.<br>
A2 exists in T3, T4.<br>
A3 exists in T1, T2, T4.<br>
A1 issued I1.<br>
A2 issued I2.<br>
A3 issued I3. <br>

Here, I would end up having it simplified to:<br>
U1 can see I1, I2, I3.

Follow-up to Question #1:<br>

I receive a payload that the ERP detached A1 and A3 from T1.<br>
I remove A1->T1, A3->T1 relationships.<br>
I need to recalculate [:CAN_SEE] since U1 loses [:CAN_SEE] to I1 because U1 to A1 has no longer valid territory relations. U1 still can see I3 because A3 loses T1 but T2 ensures the chain to I3.

How would the query look like?

Follow-up to Question #2:<br>

Active Directory got updated:<br>
U1 is detached from T1.
U1 is no longer able to see I1.

How would the query look like in this situation.

Putting my questions in another words:

I know how to create new relationships based on existing ones. I have trouble understanding how to re-invalidate old ones and remove those if no longer valid.

if I get the questions right,

1. you create :CAN_SEE relations with this query

   match (u:User)-->(:Territory)<--(:Account)-->(i:Invoice)
   merge (u)-[:CAN_SEE]->(i)

MERGE should prevent creation of multiple relations

2. & 3. if I get you right, in both case, you want to check that the path from user to invoice still exists and delete the CAN_SEE relation if the path is broken. the following query should do that

   match (u:User)-[r:CAN_SEE]-(i:Invoice) where not exists((u)-->(:Territory)<--(:Account)-->(i))
   delete r

do test before applying in production though, I did not test

I assume that your questions are for updating a graph that already contains the appropriate CAN_SEE relationships.

1. Use the following query to update the graph when a specific account (say, A1) is deleted. The query uses DETACH DELETE to delete matching paths and all relationships (including BELONGS_TO and CAN_SEE) for all deleted nodes.

   MATCH p=(a:Account)-[:ISSUED]->(:Invoice)-[:HAS]->(:InvoiceEntry)
   WHERE a.id = 'A2'
   DETACH DELETE p
   

2. Use this query to disconnect a user (U1) from a territory (T3), and to delete all the CAN_SEE relationships from that user to all invoices issued by all accounts in the territory:

   MATCH  (u:User)-[bt:BELONGS_TO]->(t:Territory)<-[:EXISTS_IN]-(:Account)-[:ISSUED]->(:Invoice)<-[cs:CAN_SEE]-(u)
   WHERE u.id = 'U1' AND t.id = 'T3'
   DELETE bt, cs