Lists Home |
Date Index |
Irene Polikoff wrote:
> Yes, modeling wife as a subclass of female may not be the right thing to
> do. Another option in either RDF or OWL is for wife or husband to be
> modeled as a property of male/female. It could then be said (using, for
> example, domain - range restrictions of RDF) that wife's must be females
> and husband's must be males.
That's a lesson we've learned the hard way in the Object-Oriented
programming world over the past decade or two. Deep semantic subclassing is
almost always a bad idea -- it makes programs hard to maintain and update,
and I'll be that it does the same thing to RDF-based taxonomies.
As far as I've seen, most object-oriented programmers have moved from heavy
subclassing to light subclassing with more aggregation. I wonder what
implications that has for how the Semantic Web people work.
All the best,