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

Re: Content negotiation: Why always redirect from non-information resource to information resource?

From: Pierre-Antoine Champin <swlists-040405@champin.net>
Date: Wed, 27 Jan 2010 13:21:20 +0100
Message-ID: <4B602FC0.40101@champin.net>
To: Christoph LANGE <ch.lange@jacobs-university.de>
CC: Peter Ansell <ansell.peter@gmail.com>, Linked Data community <public-lod@w3.org>
On 27/01/2010 12:02, Christoph LANGE wrote:
> Dear all,
> 
>   thanks for your helpful replies.  I will summarize with a few comments.
> 
> Of course the conceptual difference between a non-information resource and an
> information resource is clear to me; I was rather concerned about how relevant
> this is in an implementation where this difference is not needed.  (OK, I see,
> one might initially be tempted to think "we don't need this" but then later
> realize that "we should have sticked to good practices", so better introduce
> it from the beginning…)
> 
> Given the FOAF example by Ross, I then wondered:  If id/me 303-redirects to
> id/me.rdf, then id/me.rdf contains triples like <id/me> foaf:name "Ian
> Davis".  Now if Ian Davis (not any external user) wanted to attach metadata to
> id/me.rdf, e.g. <id/me.rdf> todo:needs "rework" where would _he_ do it?  Is it
> advisable to put them into the id/me.rdf file as well, or should he rather put
> them somewhere else?

I guess it's up to him, but in any case, he would have to alter me.rdf
either to put the information there, or to put an rdfs:seeAlso link
between me.rdf and the resource where he put his metadata.

>  In any case, the Halpin/Presutti paper mentioned by
> Pierre-Antoine, of which I am somewhat aware, seems to give a reasonable
> answer.
> 
> @Pierre-Antoine, your 200 OK / Content-Location solution appeals to me, as it
> is actually what I initially asked for.  But given its non-support by browsers
> it makes sense to me that it should not be used.

The situation is not that bad, actually. The "only" problem I'm aware of
is about using the Content-Location URI as the *base* URI for relative
links. This can be circumvented by systematically including the base in
the entity itself, using the xml:base attribute, or the @base directive
in Turtle.

On the other hand, @base appears to be unsupported by raptor and
python-rdflib, unfortunately... :-(

  pa


> I may eventually get back to you with more specific (and more philosophical!)
> questions, as the domain I am actually mostly dealing with is the management
> of mathematical knowledge.  E.g. me taking Fermat's last theorem (having
> dc:creator "Fermat") and formalizing it in an appropriate language (e.g.
> OMDoc), linking it to a proof and to examples formalized by me (or others) in
> the same formal language, finally all having dc:creator <...#me>.  There, the
> formalizations (= information resources) tend to be more important than the
> non-information resources they originate from.  One would rather not describe
> the non-information resource in the same language, but put a link to e.g.
> DBpedia.
> 
> Cheers, and thanks again,
> 
> Christoph
> 
Received on Wednesday, 27 January 2010 12:21:53 UTC

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