Re: Role of URI and HTTP in Linked Data

On Thu, 2010-11-11 at 12:59 -0600, Pat Hayes wrote:
> On Nov 11, 2010, at 8:00 AM, David Booth wrote:
> 
> > On Thu, 2010-11-11 at 07:23 +0100, Jiří Procházka wrote:
> > [ . . . ]
> >> I think it is flawed trying to enforce "URI == 1 thing" 
> > 
> > Exactly right.  The "URI == 1 thing" notion is myth #1 in "Resource
> > Identity and Semantic Extensions: Making Sense of Ambiguity":
> > http://dbooth.org/2010/ambiguity/paper.html#myth1 
> > It is a good *goal*, but it is inherently unachievable. 
> 
> Well, no, careful. It is unachievable IF the only means we have to pin
> down a referent is to describe it. 

Okay, I should have said it is inherently unachievable in the vast
majority of cases.  :)

> But that isn't the only way we have. 
> 
> In the 'real' world of ordinary language we often have the possibility
> of ostention. Q: "What do you mean by 'froogle'?"  A: "This, here in
> my hand, this is a froogle"  spoken while brandishing a teaspoon, say,
> leads to the hypothesis that 'froogle' means teaspoon, or at any rate
> some category which overlaps teaspoons. Arguably, this is at the root
> of how we learn language in the first place. (After all, if Quine were
> right about the indeterminacy of translation - 'gavagai' and all that
> - then it would be *impossible* to learn a language; and yet we all do
> it.) 
> 
> What http-range-14 does can be seen as allowing conventional HTTP
> GETting to be a form of ostention. If an http: IRI succeeds in GETting
> a representation of something with a 200 code attached, then that's
> the Web saying, in effect, A: "What I mean by <IRI> is this thing
> here"  while it - the Web - is doing the closest it can GET to
> brandishing something, ie delivering you a Web awww:representation of
> it. 

Hmm, that's an interesting view.  I kind of like it.

> (The question "What do you mean by <IRI>?" was your GET request, by
> the way, and a 303 response is the Web saying "I have no idea.")

Actually, a 303 response would be the Web saying "I don't know, but that
resource over there might have information about it".
> 
> Since the identification performed by HTTP GET is indeed unambiguous
> and unique, just as holding and waving a teaspoon is, this really does
> succeed in unambiguously identifying something. [ . . .]

Hold on, I think you're overreaching here.  I don't think the teaspoon
was unambiguously identified in any absolute sense.  It may have been
unambiguous *enough* for the task at hand.  But later training may be
needed to disambiguate the froogle from 90% of other teaspoon-like
things when I need to do a task that depends on a finer distinction.  

Similarly with a web page, the degree of identification that you get
from an HTTP 200 response may be ambiguous enough for *some*
applications, but there may be other applications that need to
distinguish between the-web-page-as-intellectual-property versus
the-web-page-as-a-bit-stream-generator.  After all, they may have
totally different custodians.

In the world of RDF, I think the notion of ambiguity is only meaningful
in a relative sense -- relative to a particular application -- not
absolute.  Is the identity of the resource unambiguous *enough* for a
particular application?  That's what matters.  And what's unambiguous to
one application may be ambiguous to another application that requires
finer distinctions.


-- 
David Booth, Ph.D.
Cleveland Clinic (contractor)
http://dbooth.org/

Opinions expressed herein are those of the author and do not necessarily
reflect those of Cleveland Clinic.

Received on Wednesday, 17 November 2010 23:42:07 UTC