Re: on owl:imports (and URI ambiguity)

>[-cc owl-dev; +cc www-archive]
>
>On Fri, 2007-06-15 at 11:15 -0500, Pat Hayes wrote:
>>  [...] The
>>  Common Logic semantics does this properly, using the notion of a
>>  'network name' which has a fixed denotation in all interpretations.
>
>Aren't you arguing with TimBL in semantic-web@w3.org that
>fixed denotations in all interpretations are nonsense?
>
>Or am I confused?

What Im arguing with Tim is that we can't just assume that a name has 
a unique denotation, or that we can make it have one by making enough 
assertions using the name (in English or in OWL/RDF/CL, though we can 
probably get good enough for practical work in English, since we seem 
to right now). That is, you can't attach names to their referents 
just by using the name in text.

But it is possible to attach it to its referent by having explicit 
naming/baptism conventions, provided we can somehow 'get hold' of the 
thing we want to baptize in some language-free way, as we can for 
network resources. Its like catching someone by the collar and 
announcing, I'm going to call this guy 'Scooter'. That is part of my 
point: we need some official, standardized, naming conventions. Named 
graphs was one suggestion for an explicit naming convention. The CL 
semantics presumes that this has been done somehow, so that CL 
modules (=ontologies) have a name which is required to be a network 
identifier, i.e. to be suitably 'attached' to the module by network 
protocols. Then the semantics is that in *any* interpretation, it 
denotes what it identifies (actually a bit more complicated than this 
because it denotes the text without the name, so you can use equality 
on them.) This is just imposed by the formal semantics. (What it 
actually does is define a 'proper' interpretation as one that gets 
this right, then as a conformance condition for a logic on a network, 
redefines satisfaction w.r.t. proper interpretations.)

There has to be some actual convention like this in the spec 
somewhere, and RDF/OWL doesn't have one. Unless you build it into the 
actual language spec somewhere, you can't get the name-referent 
attachment unique just by saying lots of stuff in the language.

Pat

>
>--
>Dan Connolly, W3C http://www.w3.org/People/Connolly/


-- 
---------------------------------------------------------------------
IHMC		(850)434 8903 or (650)494 3973   home
40 South Alcaniz St.	(850)202 4416   office
Pensacola			(850)202 4440   fax
FL 32502			(850)291 0667    cell
phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes

Received on Friday, 15 June 2007 20:33:22 UTC