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

Re: ACTION-103 are all OWL 1.0 ontologies representable in RDF (ISSUE-100)

From: Alan Ruttenberg <alanruttenberg@gmail.com>
Date: Wed, 12 Mar 2008 18:16:37 -0400
Message-Id: <28704598-7971-490E-9402-A7BE0A554AD1@gmail.com>
Cc: public-owl-wg@w3.org
To: "Peter F. Patel-Schneider" <pfps@research.bell-labs.com>

To rephrase issue 100 more precisely then, instead of
"Should there be valid OWL ontologies that can not be expressed in RDF"
it should be
"Should there be valid OWL ontologies that can not be expressed as  
OWL DL ontologies in RDF graph form"

As far as Action 103 goes, the answer to this question is that in OWL  
1.0 these things do exist. It was this situation that I thought was a  
bug in OWL 1.0, and ISSUE-100 is intended to address whether this  
situation should persist in OWL 1.1

To make the situation concrete, we can say the following: There exist  
OWL-DL ontologies written in the abstract syntax that when and  
rendered as RDF, such RDF would be classified as OWL-Full if  
interpreted as an OWL ontology. Or "OWL-DL"ness is not "round- 
trippable".

Unless there is more to clarify about this, I will amend ISSUE-100 to  
use the more precise statement.

-Alan


On Mar 12, 2008, at 2:41 PM, Peter F. Patel-Schneider wrote:

>
> Oops - ACTION-103
>
> From: Peter F. Patel-Schneider <pfps@research.bell-labs.com>
> Subject: ACTION-103 are all OWL 1.0 ontologies representable in RDF
> Date: Wed, 12 Mar 2008 14:35:40 -0400 (EDT)
>
>> Executive Summary:  YES
>> Non-executive Summary:  yes, with a caveat
>> Actual situation: same as in OWL 1.1
>>
>> Recall that an OWL 1.0 ontology has a separated vocabulary if
>> 1/ each URI reference in it is at most one of a class ID, a  
>> datatype ID,
>>    an individual ID, an individual-valued property ID, a data-valued
>>    property ID, an annotation property ID, an ontology property  
>> ID, or
>>    an ontology ID;
>> 2/ it doesn't use any disallowed vocabulary (e.g., owl:inverseOf,
>>    rdf:type) in the wrong way; and
>> 3/ it doesn't use the built-in names (e.g., owl:thing) in the  
>> wrong way.
>>
>> In OWL 1.0 every OWL DL ontology can be expressed in the abstract  
>> syntax
>> and also as an RDF graph, via the transformation in
>> http://www.w3.org/TR/owl-semantics/semantics-all.html#4.1 (which does
>> not depend on a separated vocabulary).  However, such RDF graphs  
>> are not
>> "OWL DL ontologies in RDF graph form" because these must have a
>> separated vocabulary.  The OWL DL and OWL Full semantics can (and  
>> likely
>> do) diverge on these these RDF graphs (note that Theorems 1 and 2 in
>> http://www.w3.org/TR/owl-semantics/semantics-all.html#5.4 do not  
>> apply
>> to them).
>>
>> The situation in OWL 1.1 as I envision it would be the same.  OWL 1.1
>> ontologies that do not have a separated vocabulary can be translated
>> into RDF graph form, but their Full semantics may (and is likely to)
>> diverge from their DL semantics.
>>
>>
>> Peter F. Patel-Schneider
>> Bell Labs Research
>
Received on Wednesday, 12 March 2008 22:17:02 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Wednesday, 12 March 2008 22:17:04 GMT