Re: When to introduce new object properties?
Gabriela Medina <[email protected]>
| Newsgroups | gmane.comp.misc.ontology.protege.general |
|---|---|
| Message-ID | <[email protected]> |
Thanks for your useful answers!. However, I still wonder if there is any guidelines or best practices on the subject. Best regards, Gabriela ________________________________ De : Samson Tu <[email protected]> À : [email protected] Envoyé le : Mardi 3 juillet 2012 3h04 Objet : Re: [protege-discussion] When to introduce new object properties? Hi, I think, when warranted (i.e., when the desire meaning is the same), you want to share properties across multiple classes and use local "only" restrictions, instead of global domain and axioms, to express how the property should be used in a class. The advantage of this approach is that it allows you to write more general axioms (e.g., SWRL rules) or queries that are applicable across multiple classes. When necessary, use subproperties to distinguish nuances of meaning. A common mistake is to use domain axiom to model the OO notion of a "class having a property". Samson On 6/29/2012 10:16 AM, Timothy Redmond wrote: > On 06/29/2012 09:49 AM, Gabriela Medina wrote: >> >> Hello, >> >> I have a question concerning the introduction of a new property. >> Actually I would like to know if it’s a “best practice” to use an >> object property to include one class (or as few as possible) in the >> domain and in the range. Is there any guidelines to know when it's >> appropiated to reuse the existent object properties or when to create >> new ones? > > My first reaction to this question is that it is one of those things > that is hard to give a clean answer. I think that if you use a common > object property the you are saying in some sense that it is doing the > same thing for all the different types of individuals that it provides > values for. So for example with the pizza ontology, does it make sense > to have the same hasTopping property for pizzas and for cupcakes or are > these notions of toppings really quite distinct. > > A simpler cleaner point is that domain statements for a property that > takes values for several different classes becomes unwieldy. Thus > > ObjectProperty: hasTopping > > Domain: > > Annotations: rdfs:comment "The disjunctions here can get out of hand." > Cupcake > or Pizza > > > starts to be difficult to maintain. > > Maybe some serious modelers would have something to say here. > > -Timothy > _______________________________________________ protege-discussion mailing list [email protected] https://mailman.stanford.edu/mailman/listinfo/protege-discussion Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03 _______________________________________________ protege-discussion mailing list [email protected] https://mailman.stanford.edu/mailman/listinfo/protege-discussion Instructions for unsubscribing: http://protege.stanford.edu/doc/faq.html#01a.03