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

Re: Semantics of owl:unionOf vs subclass ...

From: Pat Hayes <phayes@ihmc.us>
Date: Thu, 9 Jul 2009 22:23:27 -0500
Cc: Brandon Ibach <bibach@earthlink.net>, Bijan Parsia <bparsia@cs.man.ac.uk>, public-owl-dev@w3.org
Message-Id: <D68C2AD5-B0D3-4D20-90A3-765C68A3C5C0@ihmc.us>
To: Ruth Dhanaraj <ruthdhan@gmail.com>

On Jul 9, 2009, at 10:08 PM, Ruth Dhanaraj wrote:

> There may be more than one property with domain C.... isn't it useful
> to have it named for future use? Plus, the resulting info can be
> expressed in plain RDF (without OWL)...
>

Sure. You obviously *can* define C and give it a name. But you are not  
*obliged* to. There are no rules either way. Do what you think is best.

Pat Hayes

> Ruth
>
> On Thu, Jul 9, 2009 at 6:57 PM, Brandon Ibach<bibach@earthlink.net>  
> wrote:
>> That depends on what you hope to accomplish with your ontology and  
>> why
>> you're stating a domain for your property.
>>
>> A property's domain causes an assertion of that property to trigger  
>> an
>> assertion of the property's subject as a member of the class (C, in
>> your example) given as the property's domain.  Similarly, a subclass
>> relationship (say, A subclassof C) causes an assertion of an instance
>> as a member of A to trigger an assertion of that instance as a member
>> of C.
>>
>> If the only other facts about the class C are that A and B are
>> subclasses of it, then membership in that class carries no further
>> semantics, so what have you gained?
>>
>> -Brandon :)
>>
>> On Thu, Jul 9, 2009 at 6:19 PM, Ruth Dhanaraj<ruthdhan@gmail.com>  
>> wrote:
>>> Thanks for the info! Practically speaking, there's little difference
>>> between the two, correct? If you're not concerned with excluding non
>>> members of A and B, either syntax should suffice.
>>>
>>> Ruth
>>>
>>> On Thu, Jul 9, 2009 at 2:21 AM, Bijan Parsia<bparsia@cs.man.ac.uk>  
>>> wrote:
>>>> On 8 Jul 2009, at 19:31, Ruth Dhanaraj wrote:
>>>>
>>>>> Hi,
>>>>>
>>>>> I've been trying to figure out how I would write a property and  
>>>>> say
>>>>> its domain can be of type A *or* B. The RDF primer says that
>>>>> specifying multiple domains is an AND, so that's out.
>>>>
>>>> Correct.
>>>>
>>>>> As far as I can tell, the semantics go something like this:
>>>>> A subclassof C
>>>>> B subclassof C
>>>>> = C is a superset of A u B
>>>>>
>>>>> C unionOf (A B)
>>>>> = C is A u B
>>>>>
>>>>> (then I can say that my property has domain C)
>>>>
>>>> You don't need the first two axioms when the latter is an  
>>>> equivalence axiom.
>>>>
>>>>> Is this correct? What's the recommended way to specify this?
>>>>
>>>> You can do this without introducing a new term (C). I.e., (in no  
>>>> real
>>>> syntax)
>>>>
>>>> p domain unionOf(A B)
>>>>
>>>> Some versions of the Protege 3 series would do that by default  
>>>> when you
>>>> added multiple domains (or ranges).
>>>>
>>>> Cheers,
>>>> Bijan.
>>>>
>>>
>>>
>>
>
>
>

------------------------------------------------------------
IHMC                                     (850)434 8903 or (650)494 3973
40 South Alcaniz St.           (850)202 4416   office
Pensacola                            (850)202 4440   fax
FL 32502                              (850)291 0667   mobile
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
Received on Friday, 10 July 2009 03:25:25 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:57 GMT