Re: URIs: resources and contradictions was: Re: httpRange proposed text

Jonathan Borden wrote,
> Yes I would say that every time we use a single URI we refer to the
> same resource, but perhaps in different contexts. The meaning of a
> URI is the same as the meaning of the resource it refers to.
>
> How such a meaning of a resource is determined is context dependent. A
> resource may have different meanings in different contexts (e.g. as
> asserted by two different individuals).

Right. But look how closing the gap between URI and resource (first para 
above) opens up a new gap between resource and its "meaning" (second 
para). You're squeezing the semantic balloon in in one place only to 
have it bulge out in another.

If you want to say that URIs are logical constants in a uninterpreted 
formal system, then fine: that gets rid of ambiguity there. But to be 
relevant we need an interpretation (or a context if you'd prefer) which 
pins denotations of identifiers and extensions of predicates down to 
something tangible (cars or documents). And without a unique, 
cannonical, unequivocally intended interpretation, it's not possible to 
guarantee that interpreted uses of identifiers will be unambiguous, 
because the users might not share the same interpretation.

> We may choose to believe either one or the other context, that is we
> may choose to believe either one or the other set of assertions about
> a resource. That we might do this doesn't change the fact that the
> resource is identified by a URI (under RDF and the RDF model theory
> http://www.w3.org/TR/rdf-mt).

Believing "one or the other context" is opting for one or other 
interpretation of an uninterpreted system. That affects the truth 
values of assertions by affecting the denotations of identifiers and 
the extensions of predicates. Given the way you've set this up, your 
resources look more like placeholders than actual referents, so I guess 
they're unaffected. But the "meanings" of those resources most 
certainly will be, and those are what's important (they're what 
determine, ultimately, whether we're talking about cars or documents). 
And without a unique, cannonical, unequivocally intended 
interpretation, that's enough to get us ambiguity.

> "context" isn't a feature of RDF 1, but I think it's important.

This I agree with, wholeheartedly. But I can't help wondering why _you_ 
think it's necessary: context is important for disambiguation, but if 
there's no ambiguity ... ?

Cheers,


Miles

Received on Saturday, 3 August 2002 20:51:27 UTC