Re: [Fwd: RE: "information resource"]

On Thursday, October 14, 2004, 3:40:00 PM, Stuart wrote:


SW> [shifting discussion to www-tag rather than the comments archive]

SW> TAG members,

SW> In his response [1]  to our revised definition of the term "Information
SW> Resource" Patrick makes it clear that he is not satisfied with the 
SW> change that we propose.

SW> Short version
SW> --------------
SW> I think that we have two key questions to answer (hopefully at our 
SW> telcon on Monday):

SW> 1) Do we accept the first of Patricks suggested change from:
SW>        the set of  resources for which "...all of their essential 
SW> characteristics can be conveyed in a message."
SW>     to:
SW>        "An "information resource" is a resource which constitutes a body
SW> of information."?

To my mind no. This undoes all the careful work we did in Basel.
Currently, we have the notion of something whose entire essence is
digitally conveyable (eg a particular edition of an etext) and something
which clearly has information, but whose essence can only be measured or
approximated without conveying its entirety (a dog, a book in the
abstract without mentioning edition or translation).

To take an example, a resource for my fictional dog might return as a
representation its veterinary records (blood test results and so on) -
clearly a body of information, and clearly not conveying the entire
essence of the dog.

However, if the resource were described as 'vet records for fido' then
that would be conveying the complete essence.

SW> 2) Do we define an additional term for resources that are web accessible
SW> (that can be interacted with via an exchange of representations) - 
SW> Patricks proposal being for the term "Web Resource"?

I was previously in favor of this, but in Basel shifted by previous
position so that the newly redefined Info Resource was as good to me as
the previous Web Resource, without having to be troubled by whether it
was temporarily offline, currently reachable, had ever been online and
so forth.

SW> Long version
SW> --------------
SW> Patrick offers a counter proposal  (near the enf of [1]) that retains
SW> pieces of our revised definition that he likes.

SW> Substantive changes that Patrick suggests is to define terms for *both*
SW> "Web Resources" and "Information Resources".

That might work. Depends on the definitions of the two terms and which
of them we are usually talking about when discussing resources in
webarch.

SW> With respect to Patrick's offered defn of "Information Resources" AFAICT
SW> the substantive change to the wording is to speak of  information 
SW> resources as "...a resource which constitutes a body of information."
SW> rather than as the set of  resources for which "...all of their 
SW> essential characteristics can be conveyed in a message."  That is a 
SW> change I could support.

SW> With respect to the introduction of  the terms "Web resource", defined
SW> as  'A "web resource" is a resource which has one or more web accessible
SW> representations.' I can live with there being a terms for such things,
SW> and modulo the concern expressed at our F2F about using up 'web' prefix
SW> "Web resource" would do for me.

SW> A point of potential disagreement, which probaby gets to the heart of
SW> the matter is Patrick's assertion that under these definitions, not all
SW> "web resources" are "information resources" - meaning, I think, that a
SW> 'dog' is not an "information resource" but that it can have a "web 
SW> accessible" representation. That needs some thought.... but if we're
SW> talking generally of URI I think that's ok - I could conceive of a 
SW> people: URI scheme were URI designate people and I could organise for
SW> web accesible representations of such designated people to be available
SW> (albeit low fidelity and  somewhat dated).

SW> [Aside: A thing that continues to trouble me, and Patricks narrative
SW> triggers it again for me is a some shifting around as to whether or when
SW> URIs denote, designate, identify or  name. These various words get used
SW> sometimes as synonyms and at other times intentionally with differing
SW> nuiance - I'd rather we kept the number of different 'referring' words
SW> that we use in Webarch - small, and that if we do use multiple of these
SW> words that we are very clear about the differing nuiance the use of 
SW> different words is intended to convey. Except in the context of 
SW> namespace names I think Webarch uses the word 'identify' throughout.
SW> Partick's proposed text speaks repeatedly of things 'named' by URI and
SW> on one occasion of a thing 'denoted' by a URI. I would rather avoid 
SW> introducing "named" and "denoted" unless absolutely necessary - and if
SW> necessary - I will insist that we explain the circumstances where we
SW> choose one of these words (identify, name,  denote and for good measure
SW> 'designate' (used in many URI scheme specs)) over the others.]

SW> Regards

SW> Stuart
SW> --
SW> [1] 
SW> http://www.w3.org/mid/1E4A0AC134884349A21955574A90A7A56471B7@trebe051.ntc.nokia.com





-- 
 Chris Lilley                    mailto:chris@w3.org
 Chair, W3C SVG Working Group
 Member, W3C Technical Architecture Group

Received on Thursday, 14 October 2004 16:23:59 UTC