W3C home > Mailing lists > Public > www-tag@w3.org > March 2012

RE: Understanding URI Hosting Practice as Support for Documentation Discovery: 'meaning of meaning'

From: David Booth <david@dbooth.org>
Date: Wed, 29 Feb 2012 20:37:57 -0500
To: Larry Masinter <masinter@adobe.com>
Cc: Mark Nottingham <mnot@mnot.net>, "www-tag@w3.org" <www-tag@w3.org>
Message-ID: <1330565877.10438.77262.camel@dbooth-laptop>
On Wed, 2012-02-29 at 14:17 -0800, Larry Masinter wrote:
> [ . . . ] Until we have a common understanding of what it means to
> "convey" a "URI definition", defining the mechanism won't improve
> interoperability.

What interop problem do you have in mind?  

The interop problem that *I* care about, and the one that the LOD
community is struggling with, is that different URI owners are
essentially using different URI definition discovery protocols that
yield entirely different URI definitions, and a client has no
predictable way of knowing which protocol the URI owner followed.

For example, suppose the URI owner for http://example/toucan wishes to
convey a particular URI definition for that URI, and has configured its
server to respond to an HTTP GET request with a 200 OK response
containing an N3 RDF document that says:

  <http://example/toucan> a :Toucan ;
    :legTag "29332" ;
    :captureLocation "Southern Mexico" ;
    :captureDate "23-May-2011" ;
    :releaseDate "26-May-2011" .

Under a URI definition discovery protocol that follows httpRange-14
resolution rule "(a)" (let's call this protocol "A") the client seeking
the URI owner's intended URI definition for http://example/toucan would
obtain a URI definition that says something like the following (call
this URI definition X):

  http://example/toucan identifies an information resource
  that has a representation available from http://example/toucan .

Whereas under a different URI definition discovery protocol (call it
"B") the client seeking the URI owner's intended URI definition for
http://example/toucan would obtain a URI definition that says the
following (call this URI definition Y):

  <http://example/toucan> a :Toucan ;
    :legTag "29332" ;
    :captureLocation "Southern Mexico" ;
    :captureDate "23-May-2011" ;
    :releaseDate "26-May-2011" .

Clearly, these are different URI definitions, regardless of whether they
could "mean" the same thing under some bizarre interpretation.  

Note that the client has no automatable way of knowing whether the URI
owner of http://example/toucan was following protocol A or protocol B.
Thus, the interop problem is that the client seeking the URI owner's
intended URI definition mistakenly obtains the wrong URI definition if
it follows protocol A but the URI owner followed protocol B (or
conversely, if the client follows protocol B but the URI owner followed
protocol A).  

Note that the issue here is confusion over *which* URI definition the
URI owner intended to convey -- not a confusion about *meaning* of the
URI definition -- and this confusion is the result of the client not
knowing which URI definition discovery protocol the URI owner followed.

Unfortunately, part of the problem is that these URI definition
discovery protocols are not even officially documented anyway.  Jonathan
tried to document the possibilities here:
http://www.w3.org/2001/tag/awwsw/issue57/latest/
though that document is more a bag of options that a URI definition
discovery protocol might include than a list of fully described
protocols.


-- 
David Booth, Ph.D.
http://dbooth.org/

Opinions expressed herein are those of the author and do not necessarily
reflect those of his employer.
Received on Thursday, 1 March 2012 01:38:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:48:46 GMT