W3C home > Mailing lists > Public > public-lod@w3.org > March 2010

Re: A question - use 301 instead of 406?

From: Nathan <nathan@webr3.org>
Date: Wed, 24 Mar 2010 18:14:36 +0000
Message-ID: <4BAA568C.4020006@webr3.org>
To: Robert Sanderson <azaroth42@gmail.com>
CC: public-lod@w3.org
Robert Sanderson wrote:
> Hi Nathan,
> 
> On Wed, Mar 24, 2010 at 8:52 AM, Nathan <nathan@webr3.org> wrote:
> 
>> Robert Sanderson wrote:
>>> Secondly, the semantics of your descriptions resource are unclear.  Is it
>> an
>>> information resource or not?  Is it a conceptual set of all of the
>> formats
>>> of the descriptions of the original resource? If so, shouldn't it have
>> its
>>> own description?  If it's not that, what is it? If it is, how do you
>>> negotiate for which format you want the description of the set to be in,
>>> rather than the item from the set?
>> disagree (but also get your point and disagree in the nicest way
>> possible); neither the html document or the rdf are the description. the
>> description is a different thing entirely which is contained by either
>> the html document or the rdf document.
>>
> 
> 
> To see if I understand: you're explicitly modeling the abstract concept of
> the description as its own resource?
> 
> For example, Lord of the Rings: Tolkien's idea (a non information resource)
> would be
>     description/Tokien-LotR-Idea
> and then two information resources that contain that idea would be
>     data/Pete-Jackson-Movies
> and
>     data/Unwin-Allen-Edition-Book

I wouldn't say the idea.. and this is a totally different example aince
we are taking about three non-information-resources, not a
non-information-resource a description of that non-information-resource
and 2 documents containing that description.. but "Lord of the Rings" is
a story created by Tolkien; the group of words together regardless of
what format and which version / draft.

the english phrase "Lord of the Rings" refers to that thing (referent)
so to describe it with rdf you'd need to give it a uri, which is the
equiv of a referent.

referent: "the object or idea to which a word or phrase refers"

description/Tokien-LotR rdfs:label "Lord of the Rings" .

the unwin allen edition book, pete jackson movie, and all the other
versions are different things.. they can be named, you've named them in
the referent way, we definitely are talking about 3 distinct things.

in this case I'm dubious about which predicate I'd use to link the three
things together (primaryTopic seems wrong, hasVersion maybe..) so
negating that completely.

> Is there a significant advantage to doing this, rather than following the
> existing 303 convention and just collapsing the concept and instances
> together?

yes (imho) for multiple reasons..

because we are talking about multiple distinct things
 - the real world thing "London"
 - a particular description of london
 - multiple different representations of that description

the 303 see other points to a resource which may contain information
about said thing, because that thing doesn't have a representation which
can be transferred via http; and whilst a 303 to the rdf or html
directly isn't wrong in that sense; it does negate the possibility of
retrieving any information about the description, and further in a way
redefines the conceptual mapping identified by the uri for london.

let's say you've described London, and I want to refer to that
description, how do I do it? seeAlso "Robert Sanderson's description of
London".. I can only use the uri of one of the representations directly,
thus removing the possibility for content negotiation by anybody
following the seeAlso link.

further still; in a real world scenario it's entirely feasible to want a
system which creates multiple representations from a single entity you
send (via post) say you send rdf+xml and it creates / updates an n3
version too; how would you do this if you don't have a specific uri for
the description? you can't post to a .rdf document, or to a
non-information-resource.

and as per the op of this thread; it can solve his real world problem too.

I'm not suggesting you must do it this way, but you certainly can and in
many scenarios it makes sense to have this third resource.

as for this description is semantically speaking, is it a
non-information-resource or an information-resource, or is it both? imho
it's just a resource, identified by a uri, and you don't know what it is
until you dereference it and get a response. In my own mind (for
clarity) it is a conceptual map to my description of london, i may want
to update that description, delete it, have it in available in different
formats, and when dereferenced via its http uri identifier I'd like to
give you a list of all the different representations the server holds
and possibly select or suggest a the most appropriate one for you.

regards!
Received on Wednesday, 24 March 2010 18:15:18 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:25 UTC