RE: Provenance as a first-class citizen

> -----Original Message-----
> From: Joshua Tauberer [mailto:tauberer@for.net]
> Sent: Friday, March 17, 2006 11:09 AM
> To: Geoff Chappell
> Cc: 'SWIG'
> Subject: Re: Provenance as a first-class citizen
> 
> Geoff Chappell wrote:
> > 	:Sky [rdfs:subPropertyOf :hasColor; :says :Geoff] :Blue.
> 
> If you change rdfs:subPropertyOf to some other predicate that you define
> as sort of subtracting from the meaning of the original predicate, 

Yeah, I think you're right. When using with existing vocabs, the subprop
inference could be too strong. Probably something like rdfs:superPropertyOf
would be better (or ex:modifiesProperty). That'd give you more control over
the inference -- e.g. if a subprop rule looks like this:

	infer {?b ?x ?y} from {[rdfs:subPropertyOf] ?a ?b} 
		and {?a ?x ?y};

a rule for superprop might be:

	infer {?b ?x ?y} from {[rdfs:superPropertyOf] ?a ?b} 
		and {?a ?x ?y};
		and {[ex:sourceOf] ?a ?src}
		and {[rdf:type] ?src [ex:TrustedSource]}

or whatever the relevant context is....


> then
> you can do it without the problems of my last email.  For instance, use
> :claimType instead and define (X :claimType Y) to mean "X asserts the
> same thing as Y, except within the context of some claim", then it works
> fine.


	:claimType rdfs:subPropertyOf [owl:inverseOf rdfs:subPropertyOf]

-Geoff

Received on Friday, 17 March 2006 17:43:34 UTC