range, domain: Conjunctive AND disjunctive semantics both supportable; constraints should be identical

More halfbakery here...

Forgive me for zoning in late on this. I've only just begun to have a
closer interest in RDF schema issues. I'm not sure if anyone's
articulated these two observations to the list before now (don't recall
seeing them), but apologies in advance if I'm repeating anyone.

First: By mandating conjunctive semantics for range (and domain, see
below) we can satisfy both the conjunctive mob (who can use it for
validation and inferencing) and the disjunctive mob (who can use it for
validation, at least). I think.

Suppose we have a property P and two classes, A and B. With conjunctive
semantics, can we not model...

P has a range of (all members of A and B)

	P --[rdfs:range]-> A
	P --[rdfs:range]-> B

and loosely:

P has a range of (a member of the union of A and B)

	A --[rdfs:subclassOf]-> anon:C
	B --[rdfs:subclassOf]-> anon:C
	P --[rdfs:range]-> anon:C

(give anon:C a real URI if you prefer).

Are there problems with this scheme?

Secondly: it seems to me that whatever constraints (semantics) are
eventually applied to rdfs:range should be identical to those applied to
rdfs:domain.

Argument: modulo the appearance of literal values*, there is nothing
special about the direction that the arrow on an arc points in. In other
words, for every property P we can envisage a property P' such that

	x --[P]-> y	iff	y --[P']-> x

(example: we can create "isMotherOf" as the 'inverse' of "hasMother").

Then the range of P is the domain of P', and vice versa.

jan

* And the desire to reflect more of a datatype into the RDF layer may
loosen the notion that a literal can only appear on the sharp end of an
arc.

-- 
jan grant, ILRT, University of Bristol. http://www.ilrt.bris.ac.uk/
Tel +44(0)117 9287163 Fax +44 (0)117 9287112 RFC822 jan.grant@bris.ac.uk
Usenet: The separation of content AND presentation - simultaneously.

Received on Friday, 29 September 2000 05:22:03 UTC