- From: Gary Ng <Gary.Ng@networkinference.com>
- Date: Tue, 15 Jul 2003 16:49:07 +0100
- To: "Www-Rdf-Logic@W3.Org" <www-rdf-logic@w3.org>
- Cc: "Gary Ng" <Gary.Ng@networkinference.com>
Hi there, I have two general questions regarding OWL ontologies. I couldn't find the answers from the reference docs (it is probable that I have missed it) but I have the following interpretation. So I was wondering how other feels about them. Or better, what is the intended behaviour when a tool encounters these things? 1. How should a parsing tool handle things that are declared to be rdf:Resource? a. Either as instances, <rdf:Resource rdf:ID="AResource"/> b. or as the range of an ObjectProperty? <owl:ObjectProperty rdf:ID="aProp"> <rdfs:range rdf:resource="&rdf;Resource"/> </owl:ObjectProperty> I interpret it as follows: a. This should be allowed. Since rdf:Description is allowed, and everything is an instance of rdf:Resource anyway. In this case AResource should be treated as an instance of owl:Thing. b. Strictly speaking, this should not be allowed. Since although owl:Thing is a rdf:Resource, but not all rdf:Resources are owl:Thing, and for a OWL-DL tool, this has gone out of the boundary and I would say this falls into OWL-FULL. 2. Anonymous instances vs named instances. What is expected of them? Are all anonymous instances assumed unique? Given that in OWL there is no unique name assumption, an anonymous instance could potentially be identified with any other instances. For example, take the Measurement in the reference doc. A reasoner could - by influence of other axioms - infer that the Quantity individual is sameAs the Measurement individual, except now there is no way for a client tool to determine what the reasoner has done because there is no handles to the individuals. Have I missed something? Would appreciate any comments, or point me to previous mailings regarding these issues. Thanks very much, Gary p.s. Someone may have asked this before, apologies in advance. For example, in the wine/food ontology, there is no xml:base declared, but the access URL (in owl:imports) contains the ".owl" extensions. As without xml:base, the access URL is taken as the namespace, thus simply trying to import food.owl would lead to a mismatch in all names. Are we suppose to just ignore the extensions? Or should xml:base really be declared in those files? Or of course, change the access URL? Gary Ng, PhD gary.ng@networkinference.com Senior Software Engineer Network Inference (Holdings) Ltd Tel: +44 (0) 20 7616 0717 Fax: +44 (0) 20 7616 0701
Received on Tuesday, 15 July 2003 11:51:47 UTC