Can I add a where clause to a match in cypher and filter by a relationship property that is an array?

Question

I have a linked list, in neo4j that looks something like this:

CREATE (p:Procedure {id:1})
CREATE (s1:Step {title:"Do Thing 1"})
CREATE (s2:Step {title:"Do Thing 2"})
MERGE (p)-[:FIRST_STEP {parent:[1]}]->(s1)-[:NEXT {parent:[1]}]->(s2)

Now I might create another list that contains this list, and for that to work, I'd either create a separate set of relationships with a new parent value, or I'd add the new parent id to the list of parents: eg parent[1,2].

Now, is it possible to do a match like this:

match (p:Procedure)-[rel:FIRST_STEP|NEXT*]->(steps)
WHERE p.id = 1 and 1 in rel.parent
return p, steps

I can do it if I put the constraint in the initial declaration of the relationship eg -[rel:FIRST_STEP|NEXT* {parent:1}]-> , but that doesn't allow me to do the "IN" query.

Any thoughts or direction much appreciated.

Answer 1

Are there any expected use cases that will modify the list in some way, such as inserting, rearranging, or removing nodes? And if so, are the changes to one list meant to reflect changes to the other?

If these use cases exist, and if the list changes are meant to stay in sync with each other, single relationships with a list of parent ids makes sense (though the APOC Procedures library contains graph refactoring procedures that could handle either design).

If changes to one list aren't meant to reflect in the other list, then separate relationships per parent make the most sense.

Also, as far as I can tell there aren't easy operations to subtract elements from a list (you can use "+" to add an element, but you can't use "-"). I think you'd have to use a filter() to do this, which is a little awkward. It's easier syntactically to delete relationships entirely than to remove elements from lists on relationships, though that probably won't be a driving concern for your design choice.