W3C home > Mailing lists > Public > public-owl-wg@w3.org > April 2008

Re: [Imports Task Force] revised imports spec

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Tue, 22 Apr 2008 16:33:48 -0400 (EDT)
Message-Id: <20080422.163348.25324160.pfps@research.bell-labs.com>
To: schneid@fzi.de
Cc: public-owl-wg@w3.org

Well, this is the Semantic *Web*.   There has to be some way of
specifying the location of an "other" ontology.  The simplest is that
the Imports construct itself is tells this, thus Import(URI) says to
retrieve an ontology document from URI using the Web and import that.  

The other scheme is something like Import(O) means import the ontology
named O.  Then to get the Web retrieval part, there has to be a separate
mechanism that translates from O to O's location.

peter


From: "Michael Schneider" <schneid@fzi.de>
Subject: RE: [Imports Task Force] revised imports spec
Date: Tue, 22 Apr 2008 22:22:33 +0200

> Hi Peter!
> 
> Peter F. Patel-Schneider wrote:
> 
> >Basic idea:
> >
> >Imports in OWL 2 are "by location": if an ontology O contains a
> >statement "Import( someURI )", then "someURI" specifies the location of
> >the imported ontology---that is, an OWL 2 implementation SHOULD access
> >the ontology at the location "someURI" using the standard Internet
> >protocols. 
> 
> I just dare to ask this heretic question, because I'm in the mood: Why by
> location? 
> 
> My understanding of imports is /primarily/ that they are used to define
> which axioms are in the ontology: 
> 
>   If an ontology O contains statements 
>   "Import(someURI_1)", ..., "Import(someURI_n)",
>   then the complete ontology consists of all the axioms defined in O,
>   and (recursively) all the axioms defined in all the ontologies
>   denoted by the URIs ..., someURI_i, ..."
> 
> If someURI_i has also a format which tells me how to get this ontology, then
> that's "cool", of course, and people should IMO always build their URIs in
> such a way. But I would expect to have this as a convenience, not to find
> this specified by the OWL language itself.
> 
> Or am I horribly wrong here? :)
> 
> Cheers,
> Michael
> 
Received on Tuesday, 22 April 2008 20:35:03 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 22 April 2008 20:35:04 GMT