Instance Subclasses are not used as much as our shapes hope

[sfolsom]

Do we need to rethink how we identify different types of Instances, work with data sources to use Instance subclasses, or generalize our Instance shapes for just top class validation? (FWIW, I see good reasons to validate based on subclassing.)

[sfolsom]

Needs a decision from LC about subclasses before making a change or not.

[sfolsom]

Built in targets in SHACL include class-based, subjects-of a given property, or objects-of a given property. The only way to to define a target for entities that use a particular property (e.g. bf:content or bf:carrier) with a particular value would be to use the SPARQL constraints. E.g.

sh:sparql [
a sh:SPARQLConstraint ; # This triple is optional
sh:message "Values are Instances with a bf:carrier value of http://id.loc.gov/vocabulary/carriers/nc." ;
sh:prefixes ex: ;
sh:select """
SELECT $volume
WHERE {
$volume bf:carrier ?value .
}
""" ;
] .

It seems like we should tread carefully and confirm with LC about their use of subclasses before adding SPARQL queries to the spreadsheet.