W3C home > Mailing lists > Public > public-owl-comments@w3.org > March 2009

Re: [LC Response] To Jos de Bruijn Re: a few comments about the OWL 2 drafts

From: Jos de Bruijn <debruijn@inf.unibz.it>
Date: Thu, 26 Mar 2009 17:06:28 +0100
Message-ID: <49CBA804.1090101@inf.unibz.it>
To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>
CC: public-owl-comments@w3.org
Dear Peter,

Thank you for the response. I am satisfied with most of the
answers/edits. There are a few things I do want to come back to, though:

> Direct Semantics document:
>
> The definition for datatype maps in Direct Semantics extends datatype
> maps from RDF Semantics, in particular for facets.  

Not really.  The treatment of datatypes boils down to the same thing,
but the style of definition is quite different.  In RDF, datatype maps
are partial mappings from the set of IRIs to the set of datatypes. In
OWL 2 they are defined in quite a different way. I was just wondering
why you chose a new way of defining them.

This is not a big issue, though.

> Profiles document:
> 
> As stated in the document, OWL 2 RL is designed for easy and efficient
> implementation using existing forward-chaining rule systems.  Adding
> owl:Thing or reflexive object properties needs rules that operate over
> all individuals, which goes against efficiency, and may not even be
> possible in some rule systems. 

One can avoid universal quantification by using grounding, so all rule
reasoners that can deal with the current OWL 2 RL are able to deal with
these extensions.

But I accept your argument about efficiency.


> Similarly, most rule systems are
> designed for positive ground facts which dictates against allowing
> negative property assertions.

This argument puzzles me, because there are many constructs in OWL 2 RL
that allow expressing negative information, e.g.,
IrreflexiveObjectProperty, AsymmetricObjectProperty, not to mention the
negative type information of literals. Plus, negated facts are easily
encoded as class axioms using ObjectOneOf, ObjectHasValue, and
owl:Nothing. So, leaving out these features to improve the efficiency of
OWL 2 RL reasoning.

In summary, I still don't understand why negative class and property
assertions are not allowed in the profile.
Received on Thursday, 26 March 2009 16:07:32 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 March 2009 16:07:34 GMT