W3C home > Mailing lists > Public > public-owl-dev@w3.org > January to March 2007

Re: Reflexivity and antisymmetry uses cases?

From: Evren Sirin <evren@clarkparsia.com>
Date: Mon, 22 Jan 2007 10:10:29 -0500
Message-ID: <45B4D3E5.4040307@clarkparsia.com>
To: Michael Schneider <m_schnei@gmx.de>
CC: public-owl-dev@w3.org

On 1/22/07 9:22 AM, Michael Schneider wrote:
>
> Thanks, Evren!
>
> So, I see now, that my second example for antisymmetric and 
> /ireflexive/ relations, the "locatedIn" property between geographical 
> regions, should be better modeled to actually be locally /reflexive/, 
> because every region is located within itself. This is, of course, an 
> important usecase.
>
> So, technically, this would then be specified the following way, right?
>
>   ObjectProperty(locatedIn
>     type(AntisymmetricProperty)
>     type(TransitiveProperty)
>   )
>
>   Class(GeographicalRegion partial
>     selfRestriction(locatedIn)
>   )
Yes, this is how I would write it.

Regards,
Evren
>
> Cheers,
> Michael
>
> Evren Sirin wrote:
>> On 1/20/07 9:44 AM, Michael Schneider wrote:
>>>
>>> Holger Knublauch wrote on Wed, 17 Jan 2007:
>>>
>>>>  I don't remember a lot of requests for something
>>> > like owl:SelfRestriction on our mailing lists.
>>>
>>> And AFAICS, nobody here in this thread has given an example for 
>>> SelfRestrictions, yet. I thought about it yesterday evening for a 
>>> while, but could not come up with any serious example. Well, besides 
>>> this standard toy class of "SelfLovers", wherein property "loves" is 
>>> thought to behave transitively. ;-)
>>>
>>> It's easier for me to imagine that antisymmetry and ireflexivity 
>>> could become important, because, in combination with transitivity, I 
>>> am able to more precisely model all kinds of partial orderings 
>>> between instances, like e.g. ancestor relationships between people 
>>> or events, or inclusion relationships like "locatedIn"/"containedIn" 
>>> between geographical regions.
>>>
>>> But where is the "killer application" for owl11:SelfRestriction?
>> I'm not sure if you are only referring to SelfRestriction (local 
>> reflexivity) or ReflexiveProperty (global reflexivity) in general. I 
>> think it was mentioned earlier that reflexivity is used to describe 
>> part/whole relationships [1]. I don't know if this would be the 
>> "killer application" for reflexivity but it is certainly a widely 
>> used one. If you are asking about why we would want to use 
>> SelfRestriction instead of ReflexiveProperty then I can give a 
>> real-world example explaining this:
>>
>> One might be inclined to define foaf:knows property to be a 
>> ReflexiveProperty (similar to your loves example) because you want to 
>> model that every foaf:Person knows himself/herself. However, adding 
>> this innocent looking statement to FOAF vocabulary would make concept 
>> such as foaf:Document, foaf:Organization, foaf:Project unsatisfiable 
>> and any FOAF description that describes instances of these concepts 
>> would end up to be inconsistent. The reason is simple: When you say a 
>> property is a ReflexiveProperty then every individual in the universe 
>> should have that property. This means if you have an instance of 
>> foaf:Document, say MyDocument, then we infer that MyDocument knows 
>> itself. But the domain/range restrictions on foaf:knows says only 
>> foaf:Person's can have foaf:knows property. Since foaf:Document is 
>> disjoint with foaf:Person this would be a contradiction. The right 
>> thing to do in this case would be to use SelfRestriction construct to 
>> define a local reflexivity axiom for foaf:Person.
>>
>> Regards,
>> Evren
>>
>> [1] http://www.w3.org/2001/sw/BestPractices/OEP/SimplePartWhole/
>>>
>>> Michael
>>>
>>>
>>
>
>
Received on Monday, 22 January 2007 15:10:39 GMT

This archive was generated by hypermail 2.3.1 : Wednesday, 27 March 2013 09:32:54 GMT