Re: OWL Best Practice "value sets"

Thorsten

Thanks for the comment.   I am pleased you found the note useful.

However, I don't think I agree.

I think your method is based on a misunderstanding of the semantics  
of individuals.  As I understand it, the semantic of OWL mean that  
the only way your example can be interpreted is if some of the values  
are actually SAME individuals.  Individuals cannot be split and  
cannot overlap; they are either the same or different.

If you put in allDifferents for each of the sets of individuals - the  
obvious meaning of "partitioning" - the example is unsatisfiable.

Regards

Alan

On 25 Apr 2006, at 10:02, Thorsten Liebig wrote:

> Dear Alan,
>
> I recently read into the "Representing Specified Values in OWL"
> document. The
> best practice effort is very useful and strongly needed in order to  
> help
> non-sophisticated
> users in the course of developing OWL ontologies.
>
> I have two remarks on pattern 1 (values as sets of individuals).
> Within the fourth item of your considerations you state that there is
> "no way to represent
> alternative partitionings of the same feature space". This is  
> misleading
> to a certain point.
> It is indeed possible to extend a given partitioning in a way that
> allows more fine-graded
> nuances or to merge subsets (unless you state that your initial
> partitioning is allDifferent).
> Below there is an excerpt which extends the example you give (a merged
> document exists
> at
> http://www.informatik.uni-ulm.de/ki/Liebig/owl/values-as- 
> individuals-01.owl):
>
> <owl:Class>
>    <owl:oneOf rdf:parseType="Collection">
>           <Health_Value rdf:about="#good_health"/>
>           <Health_Value rdf:about="#medium_health"/>
>           <Health_Value rdf:about="#poor_health"/>
>    </owl:oneOf>
>    <owl:equivalentClass>
>       <owl:Class>
>          <owl:oneOf rdf:parseType="Collection">
>          <owl:Thing rdf:ID="very_good_health"/>
>          <owl:Thing rdf:ID="reasonable_good_health"/>
>          <owl:Thing rdf:ID="about_medium_health"/>
>          <owl:Thing rdf:ID="somewhat_poor_health"/>
>          <owl:Thing rdf:ID="extremly_poor_health"/>
>          </owl:oneOf>
>       </owl:Class>
>    </owl:equivalentClass>
>  </owl:Class>
>
> <rdf:Description rdf:about="#medium_health">
>   <owl:sameAs rdf:resource="#about_medium_health"/>
>   <owl:differentFrom rdf:resource="#very_good_health"/>
>   <owl:differentFrom rdf:resource="#reasonable_good_health"/>
>   <owl:differentFrom rdf:resource="#somewhat_poor_health"/>
>   <owl:differentFrom rdf:resource="#extremly_poor_health"/>
> </rdf:Description>
>
> <rdf:Description rdf:about="#poor_health">
>   <owl:differentFrom rdf:resource="#very_good_health"/>
>   <owl:differentFrom rdf:resource="#reasonable_good_health"/>
> </rdf:Description>
>
> <rdf:Description rdf:about="#good_health">
>   <owl:differentFrom rdf:resource="#somewhat_poor_health"/>
>   <owl:differentFrom rdf:resource="#extremly_poor_health"/>
> </rdf:Description>
>
> <Person rdf:about="#Jim">
>   <has_health_status rdf:resource="#reasonable_good_health"/>
> </Person>
>
> Now, Jim is also a "Healthy_person".
>
> I also suggest to relax your last item since at least FaCT++ and  
> Pellet
> now offer reasoning
> with nominals (which is acceptable in most cases).
>
> Regards
> Thorsten

-----------------------
Alan Rector
Professor of Medical Informatics
School of Computer Science
University of Manchester
Manchester M13 9PL, UK
TEL +44 (0) 161 275 6149/6188
FAX +44 (0) 161 275 6204
www.cs.man.ac.uk/mig
www.clinical-esciences.org
www.co-ode.org

Received on Thursday, 27 April 2006 07:28:51 UTC