RE: questions about rdfs:Datatype [Was: RE: Seeking normative definition of datatyping]

At 09:20 03/12/2002 -0500, Art.Barstow@nokia.com wrote:
>Hi Brian,
>
> > From: ext Brian McBride [mailto:bwm@hplb.hpl.hp.com]
> > Do let us know if you find this too confusing.
>
>Since the Primer is 100% Informative, it has equivalent weight
>as a posting to rec.humor.  That is, one really never knows if
>anything (i.e. the examples) in the Primer should be taken as
>fact (and thus copied).

I'm not sure what you are looking for here Art.  The primer is (we hope) 
informative.  I think you can rely on the examples it provides.  We 
certainly haven't put in any that we know to be wrong.


>Anyhow, for starters, the definition of rdfs:Datatype in the Vocab doc
>seems too sparse.

Yes, I agree.  We are working on that.

>  What is the meaning of:
>
>[[
>rdfs:Datatype - represents those resources that are RDF datatypes.
>]]
>
>What is this class used for?

Its just a way of saying that some resource is a datatype.

>   How does one implement support for this?

I'm not sure how to answer that, as I don't know what sort of processor you 
are creating.

>What is its relationship to the rdf:datatype attribute?

The rdf:datatype attribute takes a value which is the uri of a 
datatype.  An rdf/xml parser will spit out a statement containing a literal 
of the form:

   a b "foo"^^datatype .

which entails

   datatype rdf:type rdfs:Datatype .


>Concepts says:
>
>[[
>RDF provides no mechanism for defining new datatypes. XML Schema
>Datatypes [XML-SCHEMA2] provides an extensibility framework suitable
>for defining new datatypes for use in RDF.
>]]
>
>I think it would be very useful if a normative example or description
>for doing the above was included.

Noted.


>BTW, the links in [1] do lead to datatyping info but the primer's
>Typed Literals section doesn't appear to include any of those links.
>
> > What the WG has done is to define how to represent datatyped
> > values in RDF,
> > given there is a datatype meeting the constraints that RDF
> > imposes defined
> > somewhere.  One of the things we have not done is to provide an RDF
> > mechanism for defining new datatypes.
> > So at present there is no mechanism to use rdfs:Datatype to
> > define new
> > datatypes nor to describe the lexical space of a datatype.
>
>I wasn't able to parse the first sentence but if rdfs:Datatype
>is not used to define datatypes (I can't imagine why I thought
>it would be used for that purpose:-)), then the Vocab spec should
>clearly state that.

I guess its no big deal to say that, but I'm not following why you might 
have thought it was used to define a datatype.  I've probably been immersed 
in this stuff too long, but can say why it is that you thought it ought to be.

What set any expectation that rdfcore would provide a mechanism for 
defining new datatypes at all?


>I'm looking for a normative (in-band) solution for two problems:
>
>1. How to state a property's values may only contain a fixed set
>  of strings.
>
>2. How to state the values of a container may only contain
>  strings from a fixed set.
>
>I presmue #1 is addressed by defining an appropriate datatype using
>XML schema.  Is that correct?

Yes, then use a range constraint.  The issue will be whether your rdf 
processor will 'understand' that datatype and be able to verify the 
constraint is satisfied.


>I didn't see a solution for #2.  Do the new specs address this problem?

Sorry, no.

   http://www.w3.org/2000/03/rdf-tracking/#rdfs-constraining-containers

which states:

[[


     * Expressing such a constraint is beyond the scope of RDFS. Such 
functionality belongs with more powerful ontology languages such as 
daml+oil and owl.
* The WG notes that DAML+OIL can express this constraint as described here.
]]

where here is:

     http://lists.w3.org/Archives/Public/w3c-rdfcore-wg/2002Apr/0106.html

Hope this helps.

Brian

Received on Tuesday, 3 December 2002 10:59:52 UTC