- From: Dan Connolly <connolly@w3.org>
- Date: Tue, 31 Jul 2007 17:40:32 -0500
- To: Jeremy Carroll <jjc@hpl.hp.com>
- Cc: Owl Dev <public-owl-dev@w3.org>
On Tue, 2007-07-31 at 23:22 +0100, Jeremy Carroll wrote: > > Dan Connolly wrote: > > According to http://www.w3.org/Submission/owl11-rdf_mapping/ > > > > that seems to map to > > > > (parent brother) rdfs:subPropertyOf uncle. > > > > Meanwhile, the domain of rdfs:subPropertyOf is Property. > > So (parent brother) is a Property? I can't see any critical > > problems just now, but it seems a little odd. > > How so? > Making the bnode explicit: > > _:p rdfs:subPropertyOf uncle. > _:p = (parent brother). > > There exist a thing (_:p) that is a subPropertyOf uncle (and hence it is > a property), which equals parent composed with brother. > > Doesn't that read OK? Well, no. I can't put my finger on any sharp mathematical reasons why not, but it doesn't. It seems convenient to keep lists and properties disjoint, i.e. to remain consistent with such a constraint. It has the feel of an implicit multiplication operator a la y=mx+b Why should a list of properties be regarded as their composition, rather than their intersection or union? It certainly doesn't follow the pattern of Man owl:intersectionOf (Person Male). Adult owl:unionOf (Man Woman). That pattern suggests: uncle owlnext:compositionOf (parent brother). though I think the is/of stuff is best left outside the name, hence: (parent brother) owlnext:composition uncle. or (parent brother) owlnext:chain uncle. -- Dan Connolly, W3C http://www.w3.org/People/Connolly/
Received on Tuesday, 31 July 2007 22:40:42 UTC