Add sh:subSetOf (similar to sh:equals but for subsets)

[HolgerKnublauch]

I think this has some use cases and should be added for completeness given that we already have sh:equals and sh:disjoint.

https://datashapes.org/constraints.html#SubSetOfConstraintComponent

It is a bit similar to rdfs:subPropertyOf, but as a constraint.

Now that we allow path expressions on both sides, it can be used to ensure constraints between linked objects.

Any concerns?

[afs]

Seems like a good idea and for controlled vocabularies / enumerations values quite important.

subSetOf suggests the relationship is about explicitly modeled sets (collections).

Maybe:

sh:restrictedToValuesOf is a bit long (!).

sh:oneOf
sh:valueFrom
sh:restrictedTo
sh:restrictedBy

The DASH text:
"The property that must also have all values of the given path"
I read that as a condition on the value property, i.e. the other way round to what I think is intended.

[HolgerKnublauch]

• sh:oneOf - IMHO too close to owl:oneOf which describes enumerated classes (like sh:in)
• sh:valueFrom - IMHO too close to owl:someValuesFrom, plus it should be plural like sh:valuesFrom
• sh:restrictedTo - can work for me
• sh:restrictedBy - can work for me, less than sh:restrictedTo

I asked ChatGPT, although my prompt may have been biased, and it also suggested sh:subsetOf, alongside second choices:

• sh:valuesSubsetOf
• sh:subsetOfProperty (I guess that would need to be subsetOfPath

Note in any case the question around subSet vs. subset...

I'll raise this question in the next SHACL meeting, as names are quickly a matter of taste.

[afs]

The thing to avoid is "subset". SHACL defines constraints using "For each value node" (43 times, many in text defn boxes), not collected together and then tested except in certain cases like minCount where it is a text defn box.