addition of rdf:RDFSource on the rdf: vocabulary

[pchampin]

I found myself in a situation where I need an IRI for the notion of RDF source. More specifically, I would like to use it as the range of a property.

Since this notion is normatively defined in RDF-Concepts, I think the rdf: namespace is a natural place for this, which therefore should be described in RDF-Schemas like the rest of the terms of this namespace.

NB: this would also impact RDF-Semantics by adding a few axiomatic triples in RDFS interpretations.

[domel]

How does your proposal compare to vocabularies such as VoID, PROV, DCAT and LDP?

[afs]

For clarity, please list the necessary axiomatic triples.

[pchampin]

For clarity, please list the necessary axiomatic triples.

I believe it would only require rdf:RDFSource rdf:type rdf:Class..

How does your proposal compare to vocabularies such as VoID, PROV, DCAT and LDP?

To be clear, I don't think that such relations to other vocabularies should appear in the rdf: vocabulary itself...
But off the top of my head, I would consider the following relations to be accurate:

rdf:RDFSource rdfs:subClassOf
        void:Dataset, dcat:Dataset, prov:Entity.

void:DatasetDescription rdfs:subClassOf rdf:RDFSource.
ldp:RDFSource           rdfs:subClassOf rdf:RDFSource.

[pfps]

Is there a need for the other normatively defined concepts in Concepts?

How is this related to RDF document?

[pchampin]

Is there a need for the other normatively defined concepts in Concepts?

That's a fair question. Other such terms would probably be good candidate (RDF Vocabulary and RDF Document comes to mind).

Since RDF allows one to says anything about anything, it could be argued that every normatively defined concept deserves an IRI... But that would be a lot of IRIs (rdf11-concepts has 58 dfn tags, rdf12-concepts currently has 69), many of them of rather limited utility. So I would rather use our best judgement to curate the list of defined terms that we include in the RDF vocabulary.

How is this related to RDF document?

The definition of RDF source is focused on mutability, while the definition of RDF document is focused on RDF concrete syntaxes. Their relationship to RDF graphs is also different: an RDF source's state "can be expressed as an RDF graph", while an RDF document "encodes an RDF graph".

While some things can be both an RDF source and an RDF document, I can think of examples that are one but not the other:

• the payload of an HTTP request would be an RDF document, but not (necessarily) an RDF source;
• an HTTP resource supporting content-negociation in text/turtle and application/rdf+xml would be an RDF source, but not an RDF document (as it has no inherent "format").