W3C home > Mailing lists > Public > www-tag@w3.org > June 2011

Re: Perspective on the metadata / discovery struggle

From: David Booth <david@dbooth.org>
Date: Thu, 30 Jun 2011 12:01:44 -0400
To: Jonathan Rees <jar@creativecommons.org>
Cc: www-tag@w3.org
Message-ID: <1309449704.6147.35616.camel@dbooth-laptop>
On Thu, 2011-06-30 at 10:32 -0400, Jonathan Rees wrote:
> I had a thought about the TAG definition discovery and metadata
> architecture issues that might be helpful.  Probably this is obvious, but
> it wasn't to me so I thought it was worth writing down.  This relates
> to the fact that whenever the httpRange-14 thing comes up in the TAG
> we are confused about what issue to put it under. I was inspired to
> think this over by F2F remarks of Larry's about the magnitude of
> the problem.
> There are two distinct application-level communication needs:
>   1. web metadata - when I express information about a document
>      (image, etc.) how do I say (especially in RDF) that what I am
>      talking about is content that's accessed via a particular URI, as
>      opposed to other content
>   2. definition discovery - given a vocabulary term (URI),
>      how is definition-like information for it discovered
>      (Definitions are not, in general, metadata.)
> Described in this way, the needs seem unrelated.  The first falls
> under our ISSUE-63 (metadata architecture), the second under ISSUE-57
> (definition discovery).  The first need spawned the Resource
> Description Framework and the httpRange-14 2xx rule, while the second
> spawned linked data, RDF-style fragment ids, and GET+303.
> The connection between them is that the same notation and protocol,
> namely what I've been calling dereferencable absolute URIs, has been
> advanced as a solution to both problems.  The competition creates a
> struggle.  Think of these URIs as a limited natural resource over
> which many factions are contending.  Just as a piece of real estate
> cannot be used for a wetlands and a high-rise at the same time, one of
> these URIs can't simultaneously get its meaning according to two rules
> that give different answers most of the time.
> So the competition over "linguistic real estate" itself begets a third
> problem:
>   3. to what referential use are dereferenceable absolute URIs best
>      put?
> One might then give up and say interoperability is not a good goal,
> one might try to carve up or overload the linguistic space, one might
> try to "win", one might say it's an inadequate solution for one or the
> other problem, or that it's hopeless, and so on - all the arguments
> we've heard over and over again.
> It's not enough to solve the three problems separately. There is
> no divide-and-conquer. That is what makes it so annoying: everything
> interacts.

Interesting observation.  However, it seems to me that although #2
clearly involves using the URI referentially, #1 does not *require*
using the URI referentially.  For example, instead of using
http://example/bar referentially by writing something like:

  :fred p1:authored <http://example/bar> .

to indicate that :fred authored the content accessed from
http://example/bar , one could just as well express this using a
different property like:

  :fred p2:authored "http://example/bar" .

in which the URI is used as a literal.  This seems like another way the
conflict could be avoided.

David Booth, Ph.D.

Opinions expressed herein are those of the author and do not necessarily
reflect those of his employer.
Received on Thursday, 30 June 2011 16:02:07 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:33:10 UTC