[8pt] Add a reasoning step to saturate the ontology before generation

[NSeydoux]

When saturating the ontology, the implicit predicates are made explicit, which reduces the need for some manual vocabulary extension. For instance, X rdfs:subclassOf Y is not picked up by the generator, while it implies X a rdfs:Class. Y a rdfs:Class..

[edwardsph]

There is a similar issue with properties. If a vocab has vocab:propertyName rdfs:domain vocab:aClass but doesn't have a rdf:Property then no IRI is generated for it and we have to manually extend the vocab.

https://www.w3.org/TR/rdf-schema/#ch_domain says:

A triple of the form:
P rdfs:domain C
states that P is an instance of the class rdf:Property

If this inference was used the properties would have IRIs automatically.

[pmcb55]

Yeah, for sure - there's lots of things we could do with inference, but in fact the AG does already work with rdfs:subClassOf and rdfs:subPropertyof, see

artifact-generator/src/DatasetHandler.js

Line 748 in f769c07

// We can automatically treat anything marked as a 'sub-class of' as a

and

artifact-generator/src/DatasetHandler.js

Line 787 in f769c07

.match(subject, RDFS.subPropertyOf, null)

.
Again, if specific use-cases pop up, we can address them as needed - do you have any specific example where using rdfs:domain and rdfs:range would help you Pete...?

[edwardsph]

I was discussing this with @matthieubosquet because of acp:memberAccessControl in https://github.com/solid/authorization-panel/blob/main/proposals/acp-specification/acp.ttl.
Currently the vocab has some uses of rdfProperty alongside rdfs:domain in some places such as acp:accessControl but Matthieu is planning to remove those redundancies.