From: Matthew H. <mat...@cs...> - 2006-04-11 12:10:43
|
On 11 Apr 2006, at 01:33, Ron Alford wrote: > Matthew Horridge wrote: >> All, >> >> When parsing RDF/XML, the OWLConsumer always creates >> SubPropertyAxioms >> for rdfs:subPropertyOf statements. It seems more natural for such >> statements to be represented as asserted super properties. This >> approach would also be more in line with the way rdfs:subClassOf >> statements (where the subject of the triple is a named class) are >> handled - such statements are represented as explicit superclasses >> rather than as SubClassAxioms. >> >> Would anyone object if the default representation of sub/super >> properties was changed from SubPropertyAxioms to "explicitly >> asserted" >> super properties? >> > > That actually touches on a more general issue that's annoying me - > duplication of coverage between the various Axioms and the various > 'get' > methods. I agree. From an API user's point of view, I find having to check in both places inconvenient and it also leads to more verbose and complicated code. I can understand the argument of preserving syntax to some extent, but I think that this makes the API harder to use. > Unfortunately, the Mindswap group's addition has made this > even more prevalent (OWLTransitivePropertyAxiom, and the like). > > What would suit me best is if all the various get methods became > convenience functions over the trio of OWLPropertyAxioms, > OWLClassAxioms, and OWLIndividualAxioms. This sounds reasonable to me. > Then the only information that wouldn't be a stone's throw away > from the > Ontology object are the OWL*PropertyInstances and > OWLIndividualTypeAssertions (not that I wouldn't mind them in there). > > I can detail a couple of strategies for going towards this if people > think it's a reasonable goal. Yes - I'd appreciate that. Cheers, Matthew |