W3C home > Mailing lists > Public > public-prov-wg@w3.org > February 2012

Re: Update on PROV-O OWL file (Action item 55)

From: Satya Sahoo <satya.sahoo@case.edu>
Date: Sun, 19 Feb 2012 17:03:28 -0500
Message-ID: <CAOMwk6xL32Fo7zd4A+e8Y0xNiRGuiAiWR8COACheYmmjO1BnCQ@mail.gmail.com>
To: Ivan Herman <ivan@w3.org>
Cc: Provenance Working Group WG <public-prov-wg@w3.org>
Hi Ivan,
Thanks for the clarification! We are removing "class union" constructs from
domain of all object properties in PROV-O.

Thanks.

Best,
Satya

On Fri, Feb 17, 2012 at 4:21 AM, Ivan Herman <ivan@w3.org> wrote:

>
> On Feb 16, 2012, at 21:45 , Satya Sahoo wrote:
>
> [snip]
> >
> > ie, using a union of classes as part of the domain is not allowed. The
> rules also express this. And, although a layperson in terms of hard core
> logic, I can see why: if a resource is the subject of that property, a
> simple rule engine _cannot_ find out which of the constituents of the union
> it belongs to. Ie, it cannot make any intelligent deduction.
> >
> > Can you please clarify the above point - reasoners do consistency check
> (class with no possible instance) and classification (identify inferred
> sub/super class), the above example of inferring class membership seems to
> be a RDFS entailment.
>
> Correct. Reasoners, say, DL reasoners do this stuff, but those are
> inferences on the classes themselves. However, again from my non-logician
> perspective, OWL 2 RL concentrates more on the instance data, on the
> inferences drawn on individuals. The OWL 2 RL Rules do that, and they do it
> in a way that they can be easily implemented either directed or through
> some rule engine. And that is where the problem comes in.
>
> If I have
>
> provo:hadTemporalEngine rdfs:domain A .
>
> then the rules can be used for the inference :
>
> x provo:hadTemporalEngine y . => x rdf:type A .
>
> but if I have
>
> provo:hadTemporalEngine rdfs:domain [ owl:union ( A,B,C) ] .
>
> then, again in the case of
>
> x provo:hadTemporalEngine y . => x rdf:type A .
>
> there is no rule that would help me to make any kind of statement on the
> type of 'x' v.a.v. A, B, or C. In other words, if I simply use the OWL 2 RL
> Rules, that domain statement does not provide any information that the
> reasoner could exploit on the data.
>
> I hope this is clearer (and that I am right:-)
>
> Cheers
>
> Ivan
>
>
> ----
> Ivan Herman, W3C Semantic Web Activity Lead
> Home: http://www.w3.org/People/Ivan/
> mobile: +31-641044153
> FOAF: http://www.ivan-herman.net/foaf.rdf
>
>
>
>
>
>
Received on Sunday, 19 February 2012 22:03:59 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 13:06:56 GMT