Re: Ambiguity of "RDF namespace" - was: Re: Contradicting definitions of "property"

On Wed, May 09, 2012 at 09:49:29PM +0100, Richard Cyganiak wrote:
> > I accept the analogy, but I'm unsure what you mean by "most of the
> > sequences in the namespace" not having referents.  If you mean "potential
> > sequences" then of course that is true.  But "namespace," as I see it being
> > used, implies a set of IRIs explicitly "declared" by the namespace's
> > "owner."  
> 
> Well, that's not the way I see it. I understand a namespace as the set of
> *all* IRIs starting with the namespace IRI.

That seems awfully theoretical, at odds with existing usage of "namespace",
sloppy and inconsistent though it may be.

> How about this other approach? There's no such thing as a ?namespace? in RDF.
> Just avoid the term on its own ? it's a remnant of the old XML baggage. Some
> of the other related XML terms that came along with it have proven to be
> useful though: Some RDF syntaxes offer a way of associating ?namespace
> prefixes? with ?namespace IRIs? in order to abbreviate IRIs; and there's
> ?namespace documents? which describe the terms in an RDF vocabulary all of
> whose terms start with the same IRI. The term ?namespace? on its own is just
> a sloppy way of referring either to such an RDF vocabulary or to a namespace
> IRI.

+1 - I like this!  Keep "namespace" as an adjective.  Discourage its use as a
noun by saying that "namespace" has no formal meaning in RDF and that when used
informally, usage is inconsistent.  And above all, avoid using "namespace" as a
noun in the formal specs.

> >>> Also, the current draft of RDF Concepts 1.1 [1] still says:
> >>> 
> >>>   "Vocabulary terms in the rdf: namespace are listed and described in
> >>>   detail..."
> >>> 
> >>>   This suggests that the "rdf: namespace" holds not just any IRIs, but
> >>>   "vocabulary terms" denoted by IRIs.  
> >> 
> >> That's not a valid inference. It says that *some* of the things in the rdf:
> >> namespace are vocabulary terms. It doesn't say that all of the things in the
> >> rdf: namespace are vocabulary terms.
> > 
> > Strictly speaking, I do not think the sentence says that _all_ of the things
> > in the rdf: namespace are vocabulary terms.  But it also does not say that all of 
> > the vocabulary terms in the rdf: namespace are listed and described in detail.
> 
> Not sure what your point is.

Really only that the natural-language sentence can be understood in different
ways (and mostly by people not attuned to what may be "validly inferred").  But
I see you have removed the section from Concepts 1.1 for folding into the RDF
Schema document.

> > It is not incorrect, then, to use the same IRI both as a namespace IRI
> > and to denote a vocabulary?
> 
> It's not incorrect and it is fairly common practice AFAICT.

And that's fine with me.  The confusing bit, I think, is that

    "namespace IRI" = "vocabulary IRI"

however
    
    "namespace" != "vocabulary"

> >> [[
> >> An OWL ontology is a formal description of a domain of interest, and can be
> >> used to describe the terms in an RDF vocabulary and their relationships. OWL
> >> ontologies can themselves be expressed as RDF graphs, using terms in the owl:
> >> namespace. OWL is more expressive, but also more complex, than RDF Schema. In
> >> fact, OWL contains RDF Schema as a simple subset.
> >> ]]
> > 
> > Fine, but that's an OWL ontology.
> > 
> > What about "ontology"?  
> 
> I won't go anywhere near that question.
> 
> > Does "ontology" mean "OWL ontology"?  
> 
> I suppose every OWL ontology is an ontology? But obviously there are other kinds of ontologies, and there's ontology as a branch of philosophy.
> 
> > Is [1] an ontology?  
> 
> It's a (manifestation of an) OWL ontology. Every OWL ontology is an ontology. Therefore, [1] is an ontology.
> 
> > Does its use of "owl:equivalentProperty" make it an "OWL" ontology?  
> 
> It doesn't need to use any OWL vocabulary to be an OWL ontology. As I said: ?OWL contains RDF Schema as a simple subset.?

+1 - Okay, I get it.

> (Strictly speaking, an RDF document doesn't even need to define any classes
> or properties to be an OWL ontology. *Any* RDF document is a (degenerate) OWL
> document. But that's just sort of an accident in the way OWL is defined.)

Also fine.  I do not think this necessarily belongs in RDF 1.1 but I'd love to
see it clarified _somewhere_.

Tom

-- 
Tom Baker <tom@tombaker.org>

Received on Wednesday, 9 May 2012 21:53:53 UTC