Re: making progress on httpRange-14 -- yet another suggestion

I think that we perhaps have a different space of possible solutions:

It seems like quite a few people agree this issue *could* be a problem in
certain contexts (minus Roy and those who have yet to speak). My response
to Roy inline below. Theoretically, while for some people in some contexts
it might not be a problem, for some it might - so why not provide them a
standard solution? Also, while ambiguity is everywhere, machines do not
understand ambiguity, so if the SemWeb is going to contain no mechanism
for dealing with ambiguity, then it's in trouble ala good-old fashioned
AI.
Practically, if we're using http: URIs to stand for things besides
representations in our ontologies, what do we put there? Verbal
descriptions from our RDF schema? RDF/XML served through content
negotiation? Who knows?

But there seems to be quite a range of possibilities on what one can do:

1) Using a new URI Scheme for non-information resources (tdb of Larry
Masinter, wpn, etc)
2) Keep using hashes for non-information resources (DanC suggestion)
3) Have a representation returned denoting this is a non-information
resource (human-readable XHTML EWPN, RDFS, RDF/XML through content
negotiation/URIQA)
4) Just be careful with your http URIs, and have those URIs
denoting non-information resources return 404 errors.
5) Use content-headers/MIME-types for non-info resources

> Harry Halpin says [3]
>
>  "The Semantic Web makes as its central claim that a URI is a global
>   identifier that can be shared across various boundaries. It is as
>   yet unclear when mixing the SemWeb and the OFWeb exactly how to
>   maintain that notion of identity."
>
> OK -- why do we need or want to maintain that notion of identity
> across the SemWeb/OFWeb boundary?  What potential damage is there for
> one or the other?
>
I think that the notion of identity becomes much more powerful if it is
maintained across the SemWeb/OFWeb boundary. First, it allows you to make
SemWeb statements about OFWeb data (everything from homepages to .xml
documents with URIs - i.e. what most people actually mean when they say
"metadata"). Second, if we are constricted to only making statements about
things that are in TAG-speak "non-information resources", it's unclear
HOW we could enforce that distinction to begin with. Practically, if the
SemWeb is ever going to take off it's going to be because it's integrated
with the OFWeb. There's no potential for damage if ambiguity can be
resolved
a bit more clearly than it can now.

 Roy Fielding says [4]
>
>  "There is no way that a system can determine whether a user has
>   supplied a URI for the purpose of direct, ontologically unambiguous
>   identification of a resource, or of simple indirect identification
>   via whatever URI seems most useful at the time.  That is a problem
>   of reference that will not be solved by changing the syntax of the
>   identifiers."
>
> OK -- _why_ won't "[Signalling by e.g.] changing the syntax of the
> identifiers [solve the problem]"?
>

I would have to agree with Roy on some level - yes, it's obvious Tim
Bray's blog "ongoing" is on some level *part* of Tim Bray. However, I
would apply the old B.C. Smith test of representation/implementation. X is
a representation of Y  Tim Bray (not Tim Bray) if you blow up X, Y is
still there. X is an implementation of Y if you blow up X, Y stops
existing. Thus, Tim Bray is *not* is blog, although they may be quite
intimate at times, because if I "blow up" his blog by taking his domain
away, Tim Bray will keep on living. Now, there are esoteric arguments
involving cyborgs, and
at some point perhaps we'll *all* be connected on some level to the Web
all the time, but I think that this representation/implementation
information/non-information resource distinction is still useful and
possibly confusing to machines using the SemWeb.

Received on Friday, 6 May 2005 17:30:52 UTC