TEST/SEM: FunctionalProperty, domain-of-discourse

Does:

<rdf:RDF  >
     <owl:ObjectProperty rdf:ID="prop">
       <rdfs:range rdf:resource="#Singleton"/>
     </owl:ObjectProperty>
     <rdfs:Class rdf:ID="Singleton">
       <owl:oneOf rdf:parseType="Collection">
           <rdf:Description/>
       </owl:oneOf>
     </rdfs:Class>
</rdf:RDF>

entail

<rdf:RDF  >
     <owl:FunctionalProperty rdf:ID="prop">
</rdf:RDF>

?

I haven't completely caught with the August traffic.

I have gathered that Pat has produced a semantics which tries to bridge
the gap between Peter's work and the RDF MT.

I felt that an issue that came up in the test discussion on
FunctionalProperty might help the group see advantages
in fully endorsing Pat's approach.

So in the late July discussion [1],[2],[3] of
FunctionalProperty test #003 [7] Ian was nervous about
requiring implementations to deduce that a property
is functional. Jos and I expressed neutrality.

To further enhance this question, I have added a test 004
(the one above)
which has a property whose range is a singleton set. [8]
This is necessarily functional, but is it an
owl:FunctionalProperty?

My understanding is that under Peter's semantics [4] it is so.
Why? Because the meaning of owl:FunctionalProperty under Peter's
semantics is no more and no less than the property is functional.

Pat in his semantics [5] mirrors Peter's decision that
being an owl:FuntionalProperty is exactly equivalent to being
a functional property (note the "iff" in the first table in section 2.2)
But changing that "iff" to a "if - then" construct, we change
owl:FunctionalProperty to mean that "the property is functional
and someone decided it was worth saying it!"

This appears to be what Ian requires, since it results in a model
theory that gives a non-entailment rather than an entailment.
(note: I deliberately misrepresent Ian here).

As far as I can see, Peter's semantics cannot deliver this, whereas
Pat's semantics can, quite easily. Moreover, I note that Pat's semantics
can deliver this precisely because owl:FunctionalProperty is in
the domain of discourse.

I think I agree with Ian that we should not require implementations
to be able to detect that a property is functional, and hence I
propose this as a non-entailment test.


Jeremy

[1] Ian's comments (see last part of msg)
http://lists.w3.org/Archives/Public/www-webont-wg/2002Aug/0016.html

[2] Jeremy response
http://lists.w3.org/Archives/Public/www-webont-wg/2002Aug/0025.html

[3] Jos response
http://lists.w3.org/Archives/Public/www-webont-wg/2002Aug/0039.html

[4] Peter's semantics
http://www-db.research.bell-labs.com/user/pfps/owl/semantics.html

[5] Pat's semantics
http://www.coginst.uwf.edu/users/phayes/RDFS2OWL-C.html

[7] Test #003

http://www.w3.org/2002/03owlt/FunctionalProperty/premises003
http://www.w3.org/2002/03owlt/FunctionalProperty/conclusions003

[8] Test #004

http://www.w3.org/2002/03owlt/FunctionalProperty/premises004
http://www.w3.org/2002/03owlt/FunctionalProperty/nonconclusions004

Received on Tuesday, 27 August 2002 01:03:39 UTC