Re: datatypes and RDF Schema

From: Peter F. Patel-Schneider (
Date: 10/04/01

From: Dan Connolly <>
Subject: Re: datatypes and RDF Schema
Date: Thu, 04 Oct 2001 08:52:49 -0500

> "Peter F. Patel-Schneider" wrote:
> [...]
> >                         Adding Datatypes to RDF Schema
> > 
> >                         Peter F. Patel-Schneider
> > 
> > Here is a method for adding datatypes to RDF Schema that fits well
> > with what I see as the RDF Schema philosophy.
> Yes and no... it's somewhat appealing, but it conflicts
> with a certain amount of deployed code. That's not
> necessarily a complete show-stopper, but it's something
> that the RDF Core WG is bound to give consideration to.

I would be interested to see RDF-compliant (and, yes, I know that this is
not well defined, yet) code that this proposal breaks!   (I'm not saying
that there cannot be such, of course.)


> > 2/ Allow untyped literals to denote any element of this value space for
> >    which they are lexical representations.
> any? which one? I think a precise formulation is a little
> more involved than this, but from the stuff below, I get
> the idea...

Any.  (Actually, in some sense, all, but not all at once!)  Yes the precise
formulation is somewhat involved, but this was written to be a motivational
description, not a technical one.

> > 3/ Use special URIs to refer to these value spaces and incorporate their
> >    meaning into the meaning of RDF Schema.
> > 
> > For technical details on one way to do something very close to this, see
> > the model theory for DAML+OIL at
> >
> I looked there several times during the development of DAML+OIL
> and never found the explanation
> of what's going on here. Could you tell me which text explains
> it?

The mapping for datatype classes is defined as:

Some nodes in a DAML+OIL ontology are XML Schema datatype
definitions.  Such nodes are mapped by <tt>IC</tt> into the appropriate
subset of the datatype domain.  (To recognize whether a node is an XML
Schema datatype definition it is sufficient to determine whether it is
within an XML Schema context. For the built-in XML Schema datatypes, it is
also possible to recognize them by their standard URI.)
Nodes that are instances of <tt>daml:Class</tt> are mapped into subsets of
the object domain <tt>AD</tt> by <tt>IC</tt>.
As a special case rdfs:Literal is mapped by
<tt>IC</tt> into <tt>DD</tt>.

The mapping of literals is defined as:

The <tt>IO</tt> mapping maps RDF literals into subsets of <tt>DD</tt>.
This subset must include all elements of <tt>DD</tt> that have the literal
as a lexical representation for some XML Schema datatype.
It must not include any element of <tt>DD</tt> that is in the interpretation of
any XML Schema datatype and that does not have the literal as a lexical
representation for any XML Schema datatype.
Nodes whose type is an XML Schema datatype are mapped into subsets of the
mapping of that XML Schema datatype.  
All other nodes are mapped into singleton subsets of <tt>AD</tt>.

Yes, I agree that this is not very readable, but the model theory was
written to be precise, and not necessarily very readable.  I think that the
RDFS semantics I sent out has a more readable description of the technique,
and technically simpler, to boot.


> -- 
> Dan Connolly, W3C

Peter F. Patel-Schneider

This archive was generated by hypermail 2.1.4 : 04/02/02 EST