RE: Semantics of antisymmetric properties

Giorgos Stoilos wrote on Mon, 12 Mar 2007:

> I guess the true question is "what semantics where really meant to be
> captured"? Those of asymmetric or antisymmetric properties? 

My opinion: I would prefer /antisymmetric/ properties (from "x p y" and 
"y p x" follows "x=y"). Then, to model some real asymmetric property 
like e.g. 'hasFather', I can easily add a "IrreflexiveObjectProperty" 
axiom to the ontology. On the other hand, I then do not get into 
problems with properties like 'locatedIn', where it could be ok to say 
that some 'Location' is 'locatedIn' itself. Adding a global 
'ReflexiveObjectProperty' axiom to the ontology or define some local 
'SelfRestriction' for the 'Location' class on that property would then 
be consistent.

Cheers,
Michael

> And moreover,
> which semantics are the implementations supporting at this point? Checking
> with the reasoning algorithm in the SROIQ paper I get the feeling that it
> was meant to capture antisymmetric and not asymmetric properties. Thus, it
> might be a mistake on the semantics and not on the name of them.
> 
> Best,
> G. Stoilos
> 
>> -----Original Message-----
>> From: public-owl-dev-request@w3.org [mailto:public-owl-dev-request@w3.org]
>> On Behalf Of Boris Motik
>> Sent: Thursday, March 08, 2007 11:11 AM
>> To: 'Evren Sirin'; public-owl-dev@w3.org
>> Subject: RE: Semantics of antisymmetric properties
>> 
>> 
>> Hello,
>> 
>> You are right; this is a kind of a bug. Namely, we have followed the SROIQ
>> paper, in which they say "antisymmetric", but the definition of the
>> semantics is exactly as in OWL 1.1. Probably we should change the spec to
>> call such properties asymmetric instead of antisymmetric.
>> 
>> Sincerely yours,
>> 
>> 	Boris

Received on Monday, 12 March 2007 15:24:33 UTC