- From: <Patrick.Stickler@nokia.com>
- Date: Thu, 7 Jun 2001 10:26:54 +0300
- To: sean@mysterylights.com, jborden@mediaone.net, Patrick.Stickler@nokia.com, www-rdf-interest@w3.org
- Cc: Ora.Lassila@nokia.com
> -----Original Message----- > From: ext Sean B. Palmer [mailto:sean@mysterylights.com] > Sent: 06 June, 2001 20:07 > To: jborden@mediaone.net; Patrick.Stickler@nokia.com; > www-rdf-interest@w3.org > Cc: Ora.Lassila@nokia.com > Subject: Re: What to do about namespace derived URI refs... (long) > > > > If I e.g. want to use XML Schema to provide strict > > data typing for my serialized RDF instances (e.g. strict > > date formats, enumerations of token value sets such as > > ISO language names, etc.) I need to be sure that both XML > > Schema and RDF Schema are using the same URIs to talk about > > the same things. > > Agreed, and this is a problem for most specifications of this nature, > but not XML Schema in particular, because we can be very sure what URI > to use for any particular concept. As long as we are careful to ensure > that we only use the URI References in RDF applications, and the > namespace sans "#" for XSD processors, then we have no need to worry. > As for other specifications, Jon's schema algebra is a good thing to > look at. > > DAML is a bit mixed up, seeing as how it uses the XSD namespace sans > hash to create RDF datatypes. The way we did it with EARL (after some > prodding from Dan and Aaron!) was to define our new datatypes as > explicit URIs that are a DAML union of the URIs set out in the XSD > specification. As EARL is an RDF application, there is no problem in > doing that. Had it been an XSD application, we would have done it in > the usual XSD in XML manner. But I would assert that in many cases, such as for basic data types as integer, float, string, etc. that I would rather use more established standards (e.g. IEEE, Unicode, etc.) than the XML Schema specification -- and further, if my ontology defines a specialized data type such as "more than 17", its identity should not be tied to any particular schema definition, just because I happen to be using a particular schema tool for serialization purposes. The *concept* "more than 17" is not tied to my XML Schema instance. It is part of my ontology. I might define a rdf:isDefinedBy=".../mySchema.xsd#moreThan17" but that URIref is *not* the concept itself in my ontology, which might rather be e.g. rdf:about="urn:foo:xyz/dataTypes/integer/moreThan17" etc. Again, the specification of a property or a data type in an XML Schema is not the identity of that data type for an abstract ontology. There rather should be a mechanism for defining the components of an abstract ontololgy in a schema independent (MIME content type independent) fashion, irrespective of how serialization is done -- and then use those schema independent identities when defining the serialization schema, and ensure that when the serialized instance is distilled into triples, the identities don't get munged. Patrick
Received on Thursday, 7 June 2001 03:27:07 UTC