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

Re: Conneg representation equivalence

From: Pierre-Antoine Champin <swlists-040405@champin.net>
Date: Fri, 12 Mar 2010 09:51:07 +0100
Message-ID: <4B9A007B.4080906@champin.net>
To: Linked Data community <public-lod@w3.org>
On 11/03/2010 23:55, Nathan wrote:
> Pierre-Antoine Champin wrote:
>> On 11/03/2010 11:04, Toby Inkster wrote:
>>> On Thu, 2010-03-11 at 02:24 +0000, Nathan wrote:
>>>> If I have multiple representations of a resource which I consider
>>>> equal, let's say one of each of the following: RDF+XML, RDF+N3, SVG
>>>> Then should all three representations be considered equivalent?
>>> They certainly *could* all represent the same thing. Whether they *do*
>>> represent the same thing is a judgement call.
>> Well, if they are accessible via the same URI, using content
>> negociation, then my reading of the HTTP specification is that they
>> *must be* representations of the same resource.
>> Not sure what Nathan means by "equivalent"...
> that I consider them semantically equal representations of a resource.
> for instance "the same" RDF encoded as N3 and RDF+XML.

I would rather write that they "represent the same resource", then.
Equivalence is really a matter of what you intend to do with it. An HTML
entity and an RDF/XML entity can represent the same resource, but they
may not be "equivalent" to a human reader...

Note that the notion of "RDF graph" is a tricky one here, because in
some cases, this is neither the resource being identified by the URI nor
the entity retrieved through HTTP... Take for example


This URI would identifies a person (1), represented by an RDF graph (2),
serialized in RDF/XML (3). According to HTTP terminology, (1) is the
resource and (3) is the entity representing (1), but (2) has no precise
status here...

It does not mean that the RDF graph *can not* be considered as a
resource. As a matter of fact, the above URI will probably 303 redirect
to something like


which would idenfity the foaf profile, and could resolve in RDF/XML or
Turtle depending on Conneg, but the common practice is more to jump from
the URI of the person to the URI of a *specific* representation (.rdf or
.ttl or .html).

>>>> Is it correct that all representations must have consistent fragment
>>>> identifiers in order to be considered equivalent? 
>>> A fragment identifier should not identify different things in different
>>> representations. (Though it may be unrepresented in some or all of the
>>> representations.)
>> Is that so?
>> If I recall correctly the URI RFC (no internet when writing the mail,
>> sorry), the semantics of fragments identifiers depends on the retrieved
>> content-type. So why would they *have* to identify the same thing?
>> That being said, I agree it sounds like a good practice. Especially if
>> you consider an RDF/XML and a Turtle representation of the same RDF
>> graph... If their fragment identifier were not consistent, that would be
>> a serious headache... But is this rule written somewhere?
> yeah in awww http://www.w3.org/TR/webarch/#fragid

and more precisely in

Indeed, I once read that but had forgotten about it (though I wouldn't
have dared to behave otherwise ;)
Thanks for remininding me :)

Received on Friday, 12 March 2010 08:51:47 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 7 January 2015 15:16:03 UTC