Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[p4-feedback] Multiple domain or range for a OWL:Property - Cc: Matthew

Timothy Redmond tredmond at stanford.edu
Wed Jan 4 07:50:18 PST 2012


> I want to represent that : a person (or organisation) may or may not 
> have homepage. but if some instance has home page, it must be either 
> person or organisation.

In this case the domain statement is exactly what you want.  If you say:

    ObjectProperty: hasHomepage

         Domain:
             Organization
              or Person


then you will get the exact result that you are describing.

But this example does create an alert for me. It constrains the 
hasHomepage property in a way that may not make sense in future versions 
of your ontology.  Thus if at some point in the future you add a class 
Project and decide that Projects are not people or organization but 
might have home pages then you would have to update your domain axiom to 
say:

    ObjectProperty: hasHomepage

         Domain:
             Organization
              or Person or Project


Over time the domain axiom for the homepage property might get more and 
more unwieldy.

This example is closely related to the pizza tutorial example where the 
Ice cream class became inconsistent because it had a topping but was not 
a pizza.

-Timothy

On 12/30/2011 08:36 AM, Muhammad Javed wrote:
> Hi Matthew,
>
> continuing to our previous discussion, you said...
>
> ResearchGroup  SubClassOf  isBasedIn some (University or Institute)
> i.e. (reading left to right) anything that is a ResearchGroup is based 
> in a University or Institute.
>
> If I use the same idea and apply it on hasHomepage property,
>
> Person         subclass of hasHomepage some XSD:String
> Organisation subclass of hasHomepage some XSD:String
>
> i.e. - Anything that is a Person or Organisation  has homepage some String
>
> this may not be true in all cases, a person (or organisation) may or 
> may not have home page.
>
> I want to represent that : a person (or organisation) may or may not 
> have homepage. but if some instance has home page, it must be either 
> person or organisation.
>
> Thanks for your help in advance. really appreciate it.
>
> Muhammad Javed
>
>
>
> On Fri, Dec 23, 2011 at 2:02 PM, Matthew Horridge 
> <matthew.horridge at stanford.edu <mailto:matthew.horridge at stanford.edu>> 
> wrote:
>
>     Hi Muhammad,
>
>     > So far i understood, If i add multiple domain or ranges to any
>     property, it take it as Intersection (AND) that means.... instance
>     must be rdf:type of both domains (or ranges)
>     > am i correct ?
>
>     Yes.
>
>     > how about if i say in Protege as follows
>     >
>     > (Domain) ResearchGroup          isBasedIn      (Range)
>     University or Institute
>     >
>     > will it give same semantics (as adding a superclass of
>     ResearchGroup "isBasedIn some (University or Institute)") or it
>     has different meanings
>
>     The semantics are different.  Adding a domain of ResearchGroup to
>     isBasedIn means that any individual that has an isBasedIn
>     relationship will be entailed (inferred) to be an instance of
>     ResearchGroup.  It does NOT mean that any instance of
>     ResearchGroup has an isBasedIn relationship.  With the domain
>     axiom alone there can be instances of ResearchGroup that don't
>     have any isBasedIn relationships (I assume this isn't correct).
>
>     In essence, a domain axiom is syntactic sugar for
>
>     isBasedIn some Thing  SubClassOf  ResearchGroup
>
>     i.e. (reading left to right) anything that has an isBasedIn
>     relationship is a ResearchGroup.  Contrast this with the
>     alternative that I suggested,
>
>     ResearchGroup  SubClassOf  isBasedIn some (University or Institute)
>
>     i.e. (reading left to right) anything that is a ResearchGroup is
>     based in a University or Institute.
>
>     > This idea of "addition of superclass" is not clear to me, is
>     there any document, which i can read to understand it clearly
>     regarding multiple domains/ranges.
>
>     Multiple domains and ranges are interpreted as an intersection (AND).
>
>     Although I really don't like the phrase "properties of a class",
>     this phrase is mentioned quite a lot on this mailing list, and if
>     you like, adding a superclass
>
>     "prop some Value"
>
>     to a class A specifies that prop is a property of class A with a
>     value of Value.
>
>     I suggest that you read the Pizza tutorial to get more of a feel
>     what these class expressions are and what you can do with them.  
>     You might also take a look at the OWL 2 primer, in particular,
>     http://www.w3.org/TR/owl-primer/#Property_Restrictions.
>
>     Cheers,
>
>     Matthew
>
>
>
>
>     >
>     > thanks for your help.
>     >
>     >
>     > regards
>     > Muhammad Javed
>     >
>     >
>     >
>     >
>     >
>     > On Fri, Dec 23, 2011 at 12:29 PM, Matthew Horridge
>     <matthew.horridge at stanford.edu
>     <mailto:matthew.horridge at stanford.edu>> wrote:
>     > Hi Muhammad,
>     >
>     > Given the way you have expressed what you want to say in prose,
>     it seems to me that domain and range axioms don't correspond with
>     what you want to say.  Instead you should use class expressions
>     (in particular some values from restrictions) to state the
>     relationships that you want to express.  So,....
>     >
>     > > 1 - ResearchGroup is either based in an University or in an
>     Institute.
>     > >
>     > >      (Domain) ResearchGroup                     isBasedIn    
>                      (Range)  Univeristy
>     > >      (Domain) ResearchGroup                     isBasedIn    
>                      (Range)  Institute
>     >
>     > Add a superclass of ResearchGroup:
>     >
>     > "isBasedIn some (University or Institute)"
>     >
>     > This states that every instance of ResearchGroup is related to
>     an instance of University of Institute via the isBasedIn property.
>     >
>     > > 2 - Publication OR NewsLetter has a publication year
>     > >
>     > >      (Domain) Publication        hasPublicationYear        
>     (Range) xsd:gYear
>     >
>     >
>     > Add a superclass of Publication:
>     >
>     > "hasPublicationYear some Year"
>     >
>     >
>     > >      (Domain) NewsLetter        hasPublicationYear        
>     (Range) xsd:gYear
>     >
>     > Add a superclass of Newsletter
>     >
>     > "hasPublicationYear some Year"
>     >
>     > > 3- IndustryCollaborators has affiliation with
>     IndustyOrganisation and Researchers has affiliation with University
>     > >
>     > >    (Domain) IndustryCollaborators       hasAffiliation        
>     (Range) IndustryOrganisation
>     >
>     > Add a superclass of IndustryCollaborators:
>     >
>     > "hasAffiliation some IndustryOrganisation"
>     >
>     > >    (Domain) Researchers                   hasAffiliation      
>       (Range) University
>     >
>     > Add a superclass of Researchers
>     >
>     > "hasAffilication some University"
>     >
>     > > If i simply add multiple values in domain and range on same
>     property, what semantic meaning it gives.
>     >
>     > Intersection  i.e. AND
>     >
>     > Hope this helps.
>     >
>     > Cheers,
>     >
>     > Matthew
>     > _______________________________________________
>     > p4-feedback mailing list
>     > p4-feedback at lists.stanford.edu
>     <mailto:p4-feedback at lists.stanford.edu>
>     > https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>     >
>     >
>     >
>     > --
>     > Kind regards
>     > Muhammad Javed
>     >
>     >
>     > Muhammad Javed, PhD Research Student, Centre for Next Generation
>     Localisation, School of Computing, Dublin City University, Dublin
>     9, Ireland
>     >
>     > Tel: +353 (0)1 700 6912 <tel:%2B353%20%280%291%20700%206912>    
>                Fax: +353 (0)1 700 6702
>     <tel:%2B353%20%280%291%20700%206702>
>     > Web: www.computing.dcu.ie/~mjaved
>     <http://www.computing.dcu.ie/%7Emjaved>
>     >
>     > _______________________________________________
>     > p4-feedback mailing list
>     > p4-feedback at lists.stanford.edu
>     <mailto:p4-feedback at lists.stanford.edu>
>     > https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
>     _______________________________________________
>     p4-feedback mailing list
>     p4-feedback at lists.stanford.edu <mailto:p4-feedback at lists.stanford.edu>
>     https://mailman.stanford.edu/mailman/listinfo/p4-feedback
>
>
>
>
> -- 
> Kind regards
> Muhammad Javed
> **
> *
> *Muhammad Javed*,**PhD Research Student, Centre for Next Generation 
> Localisation, School of Computing, Dublin City University, Dublin 9, 
> Ireland
> Tel:   +353 (0)1 700 6912                Fax:  +353 (0)1 700 6702
> Web: www.computing.dcu.ie/~mjaved <http://www.computing.dcu.ie/%7Emjaved>
> *
>
>
>
> _______________________________________________
> p4-feedback mailing list
> p4-feedback at lists.stanford.edu
> https://mailman.stanford.edu/mailman/listinfo/p4-feedback

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.stanford.edu/pipermail/p4-feedback/attachments/20120104/497926be/attachment-0001.html>


More information about the p4-feedback mailing list