RE: N3 rule for proposed Resource-Description header

> From: Jonathan Rees [mailto:jar@creativecommons.org]
>
> On Apr 4, 2008, at 12:45 PM, Booth, David (HP Software -
> Boston) wrote:
>
> > Descriptions linked from a Resource-Description header can allow
> > the URI owner to provide useful assertions about the resource, even
> > though they would be ancillary assertions rather than core
> > assertions.  I thought that was what page
> > http://esw.w3.org/topic/FindingResourceDescriptions
> > was seeking.
>
> Yes, but such information could be found in many ways, as you point
> out, and if you did, then it would have no relation to your URI
> declaration scenario, so the use case would have to do with something
> else, not with URI declarations.. Can you provide a more complete use
> case scenario that specifically has to do with URI declarations? I
> will need such a thing if I'm going to include URI declarations in
> some future summary document.

Sorry, I guess I misunderstood your intent in collecting these use cases.  The notion of URI declaration is not exactly a use case for Uniform Access to Descriptions.  Rather, it helps explain why one would want "to obtain descriptions of named things . . . specifically descriptions provided (or otherwise sanctioned) by whoever minted the URI".  That's why I said it is relevant.

>
> I actually have a very hard time coming up with use cases for follow-
> your-nose in general

That puzzles me, because page
http://esw.w3.org/topic/FindingResourceDescriptions
currently says: "any situation where follow-your-nose would help is one that could benefit from a description-location technique".  What non-f-y-n cases do you have in mind that would benefit from Uniform Access to Descriptions?

> - the only ones I know of are web closure,
> semantic browsing (eg tabulator), and self describing web, and none
> of these seems very compelling to me - they don't have constituencies
> saying "I'm having a hard time getting my work done, and if only
> follow-your-nose worked better, I'd be much happier".
>
> Maybe it works really well right now, and everyone's happy, and the
> use cases are invisible because there's no problem...
>
> > However, I'm now wondering if by "Uniform Access to Descriptions"
> > you really meant "uniform access to *core* assertions".  The key
> > difference is that use of a URI only implies agreement with its
> > *core* assertions -- not its ancillary assertions.  Which was your
> > intent?  And if your intent was ancillary assertions, then why does
> > it matter whether the URI owner provided them?
>
> If you look at the use cases on the page you'll see that none of them
> cares much about definitions of names or URI declarations of semantic
> web or anything of that sort. A Creative Commons license and a POWDER
> recommendation are certainly not core assertions.

Ok, got it.  So if we assume three significant roles:

 - URI owner: as defined in AWWW.

 - Statement author: someone who uses the URI in a statement to denote a resource.

 - Statement reader: someone who reads a statement that uses a URI to denote a resource.

then given these roles, a statement reader encountering a new URI in a statement may wish to learn about the resource denoted by that URI.  I can imagine the statement reader wishing to distinguish between:

 1. Core assertions

 2. Ancillary assertions that are endorsed by the URI owner

 3. Ancillary assertions that are endorsed by some other particular entity

 4. Other ancillary assertions

Of course, many other distinctions are possible.  It sounds like the goal of "Uniform Access to Descriptions" is to locate only #1 and #2 (though the statement reader may not care about the distinction between #1 and #2).  Is that correct?  I had been wondering "Why only #1 and #2?", but I think I am now understanding that you are specifically restricting the scope of the use cases to the functionality that a Link or other header could provide.  (Duh!  Apologies for being slow on the uptake.)

> [ . . . ]
> I would think that with Mark's proposal you might be happy to say
> something like Link: my-declaration.rdf; rel="http://dbooth.org/has-
> uri-declaration" ? Then it's clear that the document linked is
> intended (by the web server) to be a declaration (core assertions) of
> some URI.

No, if that Link header is part of a 200 response, then the linked assertions must not be treated as core assertions, as explained in lines 347-360 of
http://lists.w3.org/Archives/Public/public-awwsw/2008Apr/att-0003/rules.n3.txt
Does that make sense?

> With 303 this intent has not been communicated.

True, but I think it's reasonable for Semantic Web architecture to define this interpretation of 303.  A URI declaration needs a "baptism" (or adequate evidence thereof) to distinguish the creation of the URI-resource association from regular statements.  In some sense using 303 for this purpose in the SWeb would be hijacking the existing purpose, but I have the impression that 303 hasn't been much used for other purposes, and this hijacking is relatively harmless, but I don't have any first-hand evidence of its usage rate.  It's true that a new code would be cleaner though.



David Booth, Ph.D.
HP Software
+1 617 629 8881 office  |  dbooth@hp.com
http://www.hp.com/go/software

Opinions expressed herein are those of the author and do not represent the official views of HP unless explicitly stated otherwise.

Received on Saturday, 5 April 2008 00:24:32 UTC