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").

[pfps]

I propose closing this issue with no change to the documents.

[pchampin]

I propose closing this issue with no change to the documents.

that, or making it as "future work". I concur that it should not be in the way of CR.

[TallTed]

that would be a lot of IRIs

I fail to see any problem with this. If the entities are special enough to be called out by a dfn or similar, they seem special enough to be denoted by an IRI, such that future work can refine, embellish, and/or replace the description put forward today in the body of the dfn.

[pchampin]

that would be a lot of IRIs

I fail to see any problem with this. If the entities are special enough to be called out by a dfn or similar, they seem special enough to be denoted by an IRI, such that future work can refine, embellish, and/or replace the description put forward today in the body of the dfn.

You make a very good point. In fact we could restrict this to the definitions that we elected to export (w3c/rdf-concepts#215) -- still 53 definitions...

Alternatively (and this is a serendipitous idea coming from the fact that I worked recently on w3c/rdf-concepts#215), maybe what we really need is a namespace gathering all the concepts registered in webref ? Which would live in a place like https://www.w3.org/ns/webref/?

[pchampin]

Alternatively (and this is a serendipitous idea coming from the fact that I worked recently on w3c/rdf-concepts#215), maybe what we really need is a namespace gathering all the concepts registered in webref ? Which would live in a place like https://www.w3.org/ns/webref/?

and unsurprisingly, this idea came up before (even though applied to a subset of webref...):
https://github.com/w3c/webref/issues?q=rdf

[afs]

If the entities are special enough to be called out by a dfn or similar, they seem special enough to be denoted by an IRI, such that future work can refine, embellish, and/or replace the description put forward today in the body of the dfn.

That is quite "meta" - there is a qualitative difference between a general like concept of denotes and, say, rdf:RDFSource (which should not be changeable).

For some dfn, we'd want to use the dfn-ABC but for some, we should have a different name. e.g. subject.

The word "subject" is a concept but there is an existing rdf:subject.
That might lead to it better to have a separate namespace for concepts/definitions.

it should not be in the way of CR.

+1
Adding is easy - removing from a published the namespace is hard.
This all needs careful consideration.