RE: proposal to close ISSUE-102

Boris Motik wrote:

>Hello,
>
>Yes, this is correct. Although this is not a problem in view 
>of OWL 1.1: the typing triples are just hints for the parser on how to
>interpret a certain URI. If you really wanted to define terms, 
>you should declare them.

But remember that in the example of my previous mail I talked about an OWL-1./0/-DL ontology, which already /exists/ on the SemWeb. Is the author of this ontology expected to revise his ontology in order to make it behave as expected in OWL-1.1-DL? 

Well, it looks to me that in this case better /every/ OWL-1.0-DL ontology should be checked. It can always be the case that such an old ontology contains some sporadic "dangling" typing triple, i.e. an entity for which there is only a typing triple, but which does not occur in any other axiom. An example would be that some entities are intended to be further semantically specified in a second ontology which imports the first one. If an old OWL-1.0-DL ontology is not checked and contains such a "dangling typing triple", this triple will be ignored by the mapping to Functional Syntax. And so the typed entity will get lost. This would have the consequence that such an entity (its typing triple, to be precise) will not occur anymore, when the ontology is later mapped back to RDF again. So this "inverse roundtripping" from RDF to Functional Syntax and back would have the potential to damage old OWL-1.0-DL ontologies.


But, let's assume for the moment that every existing OWL-1.0-DL ontology will really get revised by its respective authors. And let's further assume that all authors of future OWL-1.1-DL ontologies will agree to always write declarations instead of typing triples.

Now, let's imagine that I am such an ontology author, who has learnt that he should declare instead of typing. So I start my next ontology with the following triples:

  C owl11:declaredAs owl:Class .
  D owl11:declaredAs owl:Class .
  p owl11:declaredAs owl:ObjectProperty .

Then, I decide to make class C a subclass of an "AllValues p.D" restriction, so I write:

  C rdfs:subClassOf _:r .
  _:r rdf:type owl:Restriction .
  _:r owl:onProperty p .
  _:r owl:allValuesFrom D .

Since I do not plan to use punning in my ontology, I remember that it should be ok to use the old "owl:Restriction" URI instead of "owl11:ObjectRestriction".

But, AFAICS, this ontology cannot be mapped to Functional Syntax! For the above restriction construct to be mapped, it is neccessary to have either OnlyOP(p) = true or OnlyDP(p) = true. But neither is the case here, since "OnlyOP(p) = true" demands that the set Type(p) contains an element. But Type(p) only contains an element, if there is some typing triple for p in the RDF graph. And my ontology above does not contain any typing triples!

So the above RDF graph is not a valid OWL-1.1-DL ontology, since it cannot be mapped to Functional Syntax. Is this intended?

Cheers,
Michael

>I explained this in the three e-mails that I earlier identified.
>
>
>Regards,
>
>	Boris
>
>> -----Original Message-----
>> From: public-owl-wg-request@w3.org 
>[mailto:public-owl-wg-request@w3.org] On Behalf Of Michael
>> Schneider
>> Sent: 13 March 2008 15:29
>> To: Boris Motik
>> Cc: public-owl-wg@w3.org
>> Subject: RE: proposal to close ISSUE-102
>> 
>> Hi Boris!
>> 
>> Boris Motik wrote:
>> 
>> >Hello,
>> >
>> >As for your first question, note that no OWL 1.1 ontology can
>> >consist solely of a property: ontologies consist of axioms, so there
>> >is no way for an ontology to contain an entity directly. An
>> >ontology can contain a declaration axiom for a property. Then, the
>> >translation of such an ontology into RDF would generate RDF triples.
>> 
>> It has been pointed out by Peter in a different mail that 
>this topic is
>> already covered by ISSUE-89. What I want to add here is that 
>I believe that
>> this is more than just a theoretical problem.
>> 
>> Imagine there is some "lightweight" OWL-1.0-DL ontology on 
>the SemWeb, which
>> defines a bag of terms in the following way: For each term 
>only its URI and
>> its "syntactical category" (whether it is a class, an object 
>property, or a
>> data property) is determined. But no further semantic 
>relationships between
>> these terms are specified, because the ontology's author 
>does not need/want
>> such restricting information.
>> 
>> An example:
>> 
>>   ex:Person rdf:type owl:Class
>>   ex:Homepage rdf:type owl:Class
>> 
>>   ex:hasAuthor rdf:type owl:ObjectPropery
>>   ex:hasHomepage rdf:type owl:ObjectProperty
>>   ex:knowsPerson rdf:type owl:ObjectProperty
>> 
>>   ex:hasGivenName rdf:type owl:DatatypeProperty
>>   ex:hasFamilyName rdf:type owl:DatatypeProperty
>> 
>> As far as I understand, this RDF graph will be translated 
>into an *empty*
>> Functional Syntax ontology by the OWL-1.1-DL RDF-to-FunctionalSyntax
>> mapping. Is this correct?
>> 
>> Cheers,
>> Michael

--
Dipl.-Inform. Michael Schneider
FZI Forschungszentrum Informatik Karlsruhe
Abtl. Information Process Engineering (IPE)
Tel  : +49-721-9654-726
Fax  : +49-721-9654-727
Email: Michael.Schneider@fzi.de
Web  : http://www.fzi.de/ipe/eng/mitarbeiter.php?id=555

FZI Forschungszentrum Informatik an der Universität Karlsruhe
Haid-und-Neu-Str. 10-14, D-76131 Karlsruhe
Tel.: +49-721-9654-0, Fax: +49-721-9654-959
Stiftung des bürgerlichen Rechts
Az: 14-0563.1 Regierungspräsidium Karlsruhe
Vorstand: Rüdiger Dillmann, Michael Flor, Jivka Ovtcharova, Rudi Studer
Vorsitzender des Kuratoriums: Ministerialdirigent Günther Leßnerkraus