Re: proposal to close Issue 5.8 Datatypes

summary: I'm uneasy about some aspects of this,
but I understand it better now, and I don't find
anything to object to.

details:

On Mon, 2002-11-25 at 16:22, Peter F. Patel-Schneider wrote:
> From: Dan Connolly <connolly@w3.org>
> Subject: Re: proposal to close Issue 5.8 Datatypes
> Date: 25 Nov 2002 15:57:23 -0600
> 
> > On Thu, 2002-11-21 at 19:12, Peter F. Patel-Schneider wrote:
> > > 
> > > I propose that we close issue 5.8 with the following resolution(s):
> > > 
> > > 1/ The exchange syntax for OWL will use RDF datatyping.
> > 
> > I'm not sure what it means to agree to that.
> > 
> > I suppose that "RDF datatyping" refers to something like
> > http://www.w3.org/TR/2002/WD-rdf-syntax-grammar-20021108/#section-Syntax-datatyped-literals
> 
> ... and the section of the RDF MT WD on datatyping

Umm... ok, but that doesn't seem relevant to the
exchange syntax for OWL.

> > but I'm not sure how this would show up in our
> > ref/guide/semantics specs.
> 
> Well, I'm not sure how this would show up in our guide document as well.
> My editor's draft of the semantics already has the appropriate wording.
> The reference document would also have to change in some way.
> 
> > > 2/ The abstract syntax for OWL will use <datatype><lexical> as the syntax
> > >    for typed data values and will only allow lexical forms that are valid
> > >    for the datatype, i.e., <xd:integer>1.5 is not a valid typed literal in
> > >    the abstract syntax.
> > 
> > It seems more clear to refer to the abstract syntax
> > as being for OWL DL, since it's not a syntax
> > for all of OWL.
> > 
> > > 3/ OWL will include all XML Schema built-in non-list simple types, using the
> > >    canonical URI reference for them.
> > 
> > I'm not sure what it means for OWL to include a type.
> > And I'm not sure why this is in the list in
> > addtion to 1/.
> 
> The formal definition of RDF datatyping, in the RDF MT WD, does not require
> any particular datatyping.  I'm proposing that OWL *does* include XML
> Schema built-in non-list simple types.

Ah; OK, I see what you mean now.


> > > 4/ OWL can use XML Schema non-list simple types defined at the top
> > >    level of an XML Schema document and given a name, by using the URI
> > >    reference constructed from the URI of the document and the local name of
> > >    the simple type.  That is, if U is the URI of an XML Schema document
> > >    that contains, 
> > >    <xsd:schema ...>
> > >      <xsd:simpleType name="foo">
> > >        <xsd:restriction base="integer">
> > >         <xsd:minInclusive value="1700">
> > >        </xsd:restriction>
> > >      </xsd:simpleType>
> > >      ...
> > >    </xsd:schema>
> > >    then the URI reference U#foo will be that datatype.
> > 
> > Hmm... it doesn't seem reasonable for our spec to say what
> > somebody else's URIref means. I'm not sure how to state
> > my objection clearly, since of course we can use OWL
> > to express constraints on the meaning of somebody
> > else's URIs...
> > 
> > But the form of specification you used doesn't work...
> > "if U is the URI of an XML schema document" is
> > an ill-formed definite description; there isn't
> > a well-defined mapping from XML Schema documents
> > to URIs.
> 
> Huh?  Are not XML Schema documents WWW documents?  Do not all WWW documents
> have URIs?

Hmm... well, with a certain reading of the specs (the
'connected XML document' reading discussed in
http://www.w3.org/2000/06/uriqa3934.html ), each XML document
has a URI; and XML Schema documents are XML documents.

But it's entirely possible to have two XML Schema documents
with the same URI, and that that associate different types with
the same name. I suppose we could treat that
in the usual "doctor, it hurts when I do that" manner,
but I'm still uneasy about how to specify this.


> > >    Implementations of OWL may choose to ignore the facets such a type.
> > 
> > I don't know what it means to agree to that either.
> 
> This is simply an acknowledgment that user-defined XML Schema simple types
> may be too complicated for any current reasoner to handle.  
> 
> > > 5/ Cardinality restrictions in the exchange syntax for OWL will use typed
> > >    literals, as in 
> > >              _:x rdf:type owl:Restriction .
> > >              _:x owl:onProperty ex:foo .
> > >              _:x owl:cardinality "5"^^xsd:decimal . 
> > > 
> > > The semantics document has been updated to reflect all this,
> > 
> > If you could refer me to specific text, I'd appreciate it.
> 
> Looking for ``datatype'' (case insensitive) should do the trick.

Looking for ``datatype'' where?

I presume you mean the editor's draft you pointed us to on 4Nov,
recently updated:
 
http://www-db.research.bell-labs.com/user/pfps/owl/semantics/syntax.html#2.1
  Fri, 22 Nov 2002 17:32:46 GMT

| If a URI reference is a datatype

er... category error, no? that should be something like

  IF a URI reference refers to a datatype

but this is in a discussion of syntax; there isn't an interpretation
at this point in the discussion by which URI references refer
to things.

It goes on to say...

| If a URI reference is a datatype, i.e., if there is a
| datatype definition retrievable using the URI reference,
| then that URI reference cannot be used as the ID for a class.

umm... so the syntax of our language depends on the state
of the web? i.e. the identifiers we can use for classes
and such depend on what you can retrieve from the web?

If that's the way you want to specify OWL DL, I guess
I can live with that, but it seems odd.

Section 5.2. OWL Interpretations looks pretty good...

| An OWL interpretation, I = < RI, EXTI, SI, LI >, over a vocabulary V,
| where V includes VRDFS, rdfs:Literal, VOWL, owl:Thing, and
| owl:Nothing, is a datatyped RDFS interpretation over D, a set of
| datatypes including all XML Schema build-in non-list simple types, and
| ...

> > > but some
> > > changes probably need to be made to the other documents.
> > 
> > -- 
> > Dan Connolly, W3C http://www.w3.org/People/Connolly/
> 
> peter
-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/

Received on Monday, 25 November 2002 18:13:00 UTC