W3C home > Mailing lists > Public > public-rif-wg@w3.org > February 2009

[RDF+OWL] impossible to coreify OWL 2 RL combinations

From: Jos de Bruijn <debruijn@inf.unibz.it>
Date: Mon, 02 Feb 2009 18:54:06 +0100
Message-ID: <4987333E.3050309@inf.unibz.it>
To: RIF WG <public-rif-wg@w3.org>
It is impossible to completely coreify the embedding of OWL 2 RL
combinations, because there is equality in that language.  A na´ve
axiomatization such as use of sameAs (as in [1]) is not sufficient,
because in combinations one can use arbitrary predicates.

For example, if the ontology says:
FunctionalProperty(p)
ClassAssertion(C a)
PropertyAssertion(p e a)
PropertyAssertion(p e b)

With the na´ve sameAs axiomatization we can derive
a[sameAs -> b] and b[rdf:type -> C]

However, if we add the RIF fact
q(a)

we cannot derive q(b) unless we have a full-blown axiomatization of
equality (which is doable because there are no function symbols in
Core).  This would, however, clearly defeat the purpose of not having
equality in Core.

So, either (a) we say that you effectively leave Core when going for OWL
2 RL or (b) we provide an embedding of an equality-free subset.
I would be in favor of (a), because I do not want to define yet another
OWL subset; I think the whole web of semantic Web languages is already
complicated enough.


Best, Jos



Received on Monday, 2 February 2009 17:54:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:34:03 GMT