Re: Some controversial triples

On Fri, 5 Oct 2001, Pat Hayes wrote:

> This is a request for guidance from the WG. There are some RDFS
> triples that some people think should be true in every rdfs
> interpretation, while other people have doubts. The MT itself can be
> phrased either way, so we human beings have to decide. They are all
> concerned with what might be called RDFS navel-gazing.
>
> 1  rdfs:ConstraintProperty rdfs:subClassOf rdf:Property .


Since you raise it...
I'm coming aroudn to the view that the rdfs:ConstraintResource / Property
mechanism doesn't manage to do what we (the old RDFS WG) hoped it would.
It is not clear to me what things are 'constraint resources'? I believe we
muddled up two ideas and called the resulting thing a 'constraint':

 - schemas as a technology to support 'validation' as per XML DTDs etc
 - certain properties/classes having machine-readable semantics

My current inclination (which I've not made time to back up with a
propsal) would be to drop that entire mechanism from RDFS.

> 2. rdfs:comment rdfs:range rdfs:Literal .
>
> 3. rdfs:label rdfs:range rdfs:Literal .

I've not digested the varous recent threads on literals so withhold
comment on that.

Dan


>
> Currently 1. is included in the rdfs closure, but the other two are
> not. However, Jos' engine and CWM do not generate 1..  Peter P-S
> thinks that the RDFS spec states 2 and 3; my own view right now is
> that that wording in the spec is better seen as a syntactic
> constraint, and shouldn't be stated as a range constraint. But I can
> be persuaded.
>
> Bear in mind that including such triples in a closure can have
> knock-on effects on other triples that might be inferrable from these
> using other closure rules.
>
> Anyone got any strong views on any of these one way or the other?
>
> Pat
>
>

Received on Friday, 5 October 2001 11:18:27 UTC