W3C home > Mailing lists > Public > public-owl-dev@w3.org > January to March 2009

Re: Clarification on owl:real sought

From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
Date: Thu, 08 Jan 2009 08:43:38 -0500 (EST)
Message-Id: <20090108.084338.126417953.pfps@research.bell-labs.com>
To: der@hplb.hpl.hp.com
Cc: public-owl-dev@w3.org

Your analysis is correct.

Supporting owl:realPlus (to pick a slightly more interesting example)
means something like allowing ontologies like: 

DataProperty(ex:wealth)
SubClassOf(ex:Person AllValuesFrom(ex:wealth owl:real))
ClassAssertion(ex:Person ex:Steve)
ClassAssertion(ex:Person ex:Bill)
PropertyAssertion(ex:wealth ex:Brian "+10000000000000000000"^^xsd:integer)
PropertyAssertion(ex:wealth ex:Bill "+INF"^^xsd:float)

In the absence of real constants, it is indeed the case that owl:real is
going to act very much like owl:rational, as far as OWL itself is
concerned.  If owl:rational is removed then owl:real is going to act
very much like xsd:decimal.

Perhaps the main reason for owl:rational and owl:real is to allow for
extensions of OWL that provide relationships between values.

peter



From: Dave Reynolds <der@hplb.hpl.hp.com>
Subject: Re: Clarification on owl:real sought
Date: Thu, 08 Jan 2009 09:50:13 +0000

> Thanks Peter.
> 
> That helps a little.
> 
> What I'm trying to understand is what it means to "support" owl:real
> either for OWL 2 RL implementations or for RIF (though this is not an
> official RIF enquiry either).
> 
> For OWL 2 RL I can see that you can, for example, have a
> DatatypeProperty with range owl:real and that individual values within
> xsd:decimal and values within owl:rational would both be compatible with
> that range.
> 
> Would I be right in assuming that if owl:rational is dropped (which is a
> Feature At Risk, and I'd personally be happy to see it go) then there
> would be little point including owl:real in OWL 2 RL since all
> expressible numbers would then be within the value space of xsd:decimal?
> 
> Dave
> -- Hewlett-Packard Limited
> Registered Office: Cain Road, Bracknell, Berks RG12 1HN
> Registered No: 690597 England
> 
> 
> Peter F. Patel-Schneider wrote:
> > [This is not an "official" response.  You might consider elevating
> this
> > to a formal comment, perhaps because you want better explanation to
> show
> > up in the documents.]
> > owl:real is slightly strange as a datatype, in that it has uncountably
> > many values in its value space.  This means that there is no way to
> have
> > elements of its lexical space for all of its values.  (Well, we could,
> I
> > suppose, but that might have some computational consequences for OWL,
> > and even for storing and parsing OWL documents. :-) )
> > I think that at one time, the OWL WG did discuss the idea of having
> > lexical elements for certain interesting owl:real values (like pi or
> the
> > square root of 2).  The current status, however, is that the lexical
> > space for owl:real is empty.  (Remember that empty is a perfectly
> > reasonable set!)
> > This does not mean that elements of the value space of owl:real cannot
> > be used in OWL ontologies.  For example, "1/3"^^owl:rational denotes
> an
> > element of the value space of owl:real (just one that also an element
> of
> > the value space of owl:rational).
> > This illustrates a subtle difference between the treatment of
> datatypes
> > in OWL and in XML Schema (but one that is described in the OWL
> > documents).
> > I hope that this answers your question.
> > Peter F. Patel-Schneider
> > From: Dave Reynolds <der@hplb.hpl.hp.com>
> > Subject: Clarification on owl:real sought
> > Date: Wed, 07 Jan 2009 11:39:17 +0000
> > 
> >> [This is not a formal comment, just seeking understanding.]
> >>
> >> In the OWL 2 Syntax specification [1] section 4 it states that every
> >> datatype in the datatype map is described by a value space, a lexical
> >> space and a facet space. In section 4.1 it lists owl:real as such a
> datatype but then says that
> >> owl:real does not "directly provide any lexical values". Could
> someone
> >> explain in what sense owl:real is a datatype if there are no lexical
> >> values? Since owl:real is included in the OWL 2 RL profile that seems
> to imply
> >> it is intended to be used in OWL 2 RL documents which would seem to
> >> imply lexical values. Dave
> >>
> >> [1] http://www.w3.org/TR/owl2-syntax/
> >> [2] http://www.w3.org/TR/owl2-profiles/#Feature_Overview_3
> >> -- Hewlett-Packard Limited
> >> Registered Office: Cain Road, Bracknell, Berks RG12 1HN
> >> Registered No: 690597 England
> >>
> >> Ivan Herman wrote:
> >>> The W3C OWL Working Group has just published seven drafts for OWL 2,
> >>> including the structural specification, direct and RDF based
> semantics,
> >>> serialization in RDF or in XML, Profiles, conformance and test
> cases. See
> >>> http://www.w3.org/blog/SW/2008/10/10/seven_owl_2_drafts_published
> >>> for more details and pointers to the documents themselves. The
> Working
> >>> group seeks public feedback on the drafts; send your comments to
> >>> public-owl-comments@w3.org. Please, send your comments until
> 2008-10-23.
> >>> Cheers
> >>> Ivan
> > 
> 
> 
Received on Thursday, 8 January 2009 13:42:13 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:56 GMT