Consider changing "concept has only one label" to skos:prefLabel has only one label per lang.

[RickMoynihan]

pmd4/src/swirrl/validations/pmd4/SELECT_ConceptExactlyOneLabel.sparql

[Robsteranium]

To be clear, I presume we want to confirm that the concept has:

• no more than one skos:prefLabel per language tag
• no more than one rdfs:label
• at least one of skos:prefLabel or rdfs:label (if both the former is preferred)

We should defer relaxing this to permit multiple (different) labels scoped by graph until we've decided how to identify which label to present in a given context (https://github.com/Swirrl/cogs-issues/issues/34).

[Robsteranium]

Have just huddled with Rick on this. We've agreed to:

• delete the pmd4/src/swirrl/validations/pmd4/SELECT_ConceptExactlyOneLabel.sparql validation
• introduce (error) validations to require:
  • "no more than one skos:prefLabel per language tag"
  • "at least one of skos:prefLabel or rdfs:label (if both the former is preferred)"
• introduce a warning to require "no more than one value of rdfs:label if skos:prefLabel isn't present" (i.e. we can permit the same label to appear in different graphs since this is inconsequential for the UI), this warning should basically instruct users to provide a skos:prefLabel or risk not having explicit control over which label PMD picks. We can later relax this to e.g. "no more than one value (...) per graph" once we decide how to scope concept descriptions by graph Swirrl/cogs-issues#34.

We ought to coordinate this with a change to Muttnik to follow this label cascade https://github.com/Swirrl/muttnik/issues/1403.

[Robsteranium]

This is implemented with d37f75866855843ebe71bc9116de994af17e0fb9.

We can downgrade ConceptMaxOneLabelIfNoPrefLabel.sparql as part of #20.