Re: [re-send] Re: XSLT: status of owls:Documentation unclear

From: Dan Brickley <danbri@w3.org>
Subject: Re: [re-send] Re: XSLT: status of owls:Documentation unclear
Date: Sun, 26 Jan 2003 15:36:24 -0500

> 
> * Masahiro Hori <HORIM@jp.ibm.com> [2003-01-27 05:29+0900]
> > 
> > 
> > Should we think about allowing additional metadata (.e.g., DC and TopicMap)
> > into arbitrary position of an OWL document?  Or, is it sufficient to allow
> > the inclusion of additional metadata in the header section of OWL?
> 
> Yes please! Some metadata is better expressed per-class, per-property; for
> eg. pointers to test cases, decisions of the creating organisation., etc etc.
> RDFS allows arbitrary additional metadata for classes and properties, which is 
> why OWL can be layered on top of RDFS. 

If only this were so.  

RDF allows (relatively) arbitrary triples to be connected to resources, but
by no means does it allows arbitrary additional metadata for classes and
properties.   Getting around the restrictions imposed by RDF has consumed
an incredible amount of time in the WebOnt working group, and only a
partial solution has been achieved for the ontology level.  Moving to full
first-order logic will require yet another time-consuming design effort.

> A similar flexibility in OWL's design 
> might allow an OWL 1.1 or OWL 2.0 to be added later...

The design of OWL is actually more flexible than the design of RDF, in that
OWL does not attempt to dictate its own meaning for all possible syntax.
This allows for future efforts that come with their own meaning for some
new syntax, provided that, and this is a gigantic provisio, this new
meaning fits within the meaning mandated by RDF.

I would hope that at some time in the near future a more-flexible design
can be devised for the Semantic Web.

> Dan

Peter F. Patel-Schneider
Bell Labs Research
Lucent Technologies

Received on Monday, 27 January 2003 09:08:59 UTC