- From: Sandro Hawke <sandro@w3.org>
- Date: Sun, 28 Jan 2007 23:18:27 -0500
- To: "Mike Dean" <mdean@bbn.com>
- Cc: public-rif-wg@w3.org
> > In this case, the "Address"
> > stripe is redundant if it's known that the range of the "shipTo"
> > property is "Address".
>
> What if the desired value for shipTo is an instance of a subclass of
> Address, e.g. HomeAddress or BusinessAddress?
Then you can't skip the stripe. (At least not with the kind of stripe
skipping I think we should use. In theory you could construct a system
where such stripes could be skipped if the class information were
derivable from property information, I think.)
> Is stripe-skipping dynamic (i.e. based on the data) or static (i.e. based on
> the ontology/schema)?
It's always based on the ontology/schema, but in the dynamic case it's
also based on the data. I mentioned <Var> inside <declare> as an
example where you could skip a stripe dynamically -- but I recommend
against RIF doing that, since it makes everything more complicated.
A related question would be: is it okay in RIF (given the asn06 in the
last e-mail) to have a <Condition> element? Or do you always have to
use the leaves of the subclass tree -- always say <Atom>, <And>,
<Exists>, or <Equals> ? My intuition is to stick with the leaves, but
there might be trees where that's not what you want.
-- Sandro
Received on Monday, 29 January 2007 04:18:56 UTC