W3C home > Mailing lists > Public > www-webont-wg@w3.org > January 2003

Re: Guide Action: add "everything in Lite needs a type" to Guide

From: Jeremy Carroll <jjc@hplb.hpl.hp.com>
Date: Tue, 28 Jan 2003 08:01:45 +0000
Message-ID: <3E3638E9.3070702@hpl.hp.com>
To: Christopher Welty <welty@us.ibm.com>
CC: www-webont-wg@w3.org



It's not so much that everything needs to be a Class, Property or 
Individual but that everything must have an explicit type.
i.e. for every URIref xxx used as a subject, property or object and every 
blank node there must be a triple

xxx rdf:type yyy .


with the following exceptions

1. yyy = rdfs:Class, rdfs:Property, owl:FunctionalProperty,
        owl:InverseFunctionalProperty, owl:TransitiveProperty,
        owl:SymmetricProperty
don't count

2.  most built-ins (owl, rdf, rdfs namespace stuff) should not be typed
   (I cannot yet clarify most)

3. properties and objects used in annotations must not have types

4. (DL only) datarange bnodes must not be typed (the ugly case)

5. (DL only) bnodes of descriptions are optionally typed

6. datatypes must be typed iff they occur in some data literal other than 
those used in owl cardinality constraints!! (wonderful that one)

7. imported ontologies are currently broken in OWL DL syntax, but with the 
obvious fixes would optionally be typed.

(possibly some other special cases)
===

I am working on a cleaned up OWL DL syntax that would, if the WG likes it, 
simplify this to 1 and 2 only (both simplified somewhat).


I suspect this comment originated with me before I had properly understood 
how bad the OWL DL/OWL Lite concrete syntax is.

Sorry, I don't believe the constraints can be turned into text appropriate 
for a Guide reader.

Jeremy



Christopher Welty wrote:

> In the editors breakout session, someone suggested that I add to Guide 
> something to the effect that, in Lite, everything needs a type.
> 
> I've been trying for the past week to figure out what that means - each 
> time I try I end up at the same place.  I tried to run it past PeterPS, 
> and he ended up there too: In Lite (and DL), everything must be either a 
> Class, Property, or Individual, and nothing more.   But the Guide already 
> says that.
> 
> I can't remember whose suggestion this was - if whoever suggested it feels 
> there is something missing from Guide, please give me a more concrete 
> suggestion (like some text to add).
> 
> -Chris
> 
> Dr. Christopher A. Welty, Knowledge Structures Group
> IBM Watson Research Center, 19 Skyline Dr.
> Hawthorne, NY  10532     USA 
> Voice: +1 914.784.7055,  IBM T/L: 863.7055
> Fax: +1 914.784.6078, Email: welty@us.ibm.com
> 
> 
Received on Tuesday, 28 January 2003 03:02:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:57:57 GMT