don't use inverse props, use "virtual inverses"

[VladimirAlexiev]

You define 2 inverses of existing props: isDistributionOf, isAccessServiceOf.
(The diagram mentions that only the latter is "inverse", but the ontology defines both as owl:inverseOf).

It's better not to introduce inverses since they are redundant, but pose additional reasoning or querying requirements.
See https://www.w3.org/TR/prov-o/#inverse-names for a justification.

It's better to use "virtual inverses" (see the above link and https://www.w3.org/TR/json-ld11/#reverse-properties) eg:

• declare the inverse name to use: dcat:distribution prov:inverse "isDistributionOf"
• add to @context:

"dprod:isDistributionOf": {"@reverse": "dcat:distribution"}

[rivettp]

I'm personally inclined against inverses for the reasons mentioned and more; however this requires wider discussion. I wish the 2 alternatives you mention had wider adoption (e.g. a non-JSON-LD equivalent).

[VladimirAlexiev]

Ontotext Semantic Objects have virtual inverses: https://platform.ontotext.com/semantic-objects/soml/properties.html#inverses. That's mandatory for GraphQL because it doesn't have the universal inversability of SPARQL.