Re: range, domain: Conjunctive AND disjunctive semantics both supportable

Ian Horrocks wrote:
> 
> On October 1, Wolfram Conen writes:
> > Dear Jan and dear RDFS folks interested in range constraint discussion,
> >
> > We will make two claims in the following:
> >
> > (C1) Disjunctive/Conjunctive range constraints is not all that would be
> > needed! General set-based range constraints should be able to constrain
> > the range to an arbitrary set-algebraic expression.
> 
> Yes, and in OIL you can do just this.
> 
> > (C2) (Exactly) one range constraints is sufficient to express any
> > set-related range contraint, if sets can be constructed from
> > set-algebraic operations.
> 
> Obviously. But the trouble is that in a web environment different
> people may add various constraints in various different places. It
> wont in general be possible to get them all to agree to withdraw their
> individual constraints in favour of a single algebraic constraint, so
> we need both multiple constraints and some semantics for how to
> interpret them.
> 

Partially. You need one constraint to an "abstract" class that represent
the range for each property (would be good modelling discipline to
define it pretty "close" to the definition of the property) -- and you
have to allow that the content of this class can be constructed
"distributedly" (would be nice, if a "directory" service could be used
to announce further constraints) -- and that is exactly what is
described in the "lower" part of the Email: how you can construct
classes from other classes with triples/reification/elementary
properties (u,n,/) in RDF/RDFS.

No one would be forced to withdraw a constraint - in fact, no one would
ever use a range constraint, because it would be expressed by class
construction. However, you are certainly right, that the open/closed
world problem remains. The construction is not necessarily monotonous -
but "such is life...nonmonotonous" - one solution could be AWARENESS (be
aware that your knowledge may be incomplete and will time out fast --
and try to be aware of constraints that other impose in the
properties/classes/resources you want to describe/use) -- and it would
be nice to support this awareness actively (giving people a possibility
to announce: hey, I have used this property p too and I have
added/subtracted/intersected a class to the range class)

Wolfram + Reinhold

Received on Monday, 2 October 2000 12:18:54 UTC