[rdfweb-dev] Fwd: uri as name or link to another rdf?

Danny Ayers danny666 at virgilio.it
Thu Jan 8 16:23:32 UTC 2004


> >> I still don't get one basic RDF issue
> >> right: How should any rdf reader (person or agent) know whether the
> >> uri "http://www.schemaweb.info/parser/license.xml" is just a name for
> >> a thing or whether you should look at that uri, do a HTTP-GET and
> >> interpret (here: parse XML, internalize RDF) what you get? I haven'd
> >> found that basic distinction.

Hmm, I guess the question is why you'd want to try a GET. In theory they are
just identifiers, but in practice you may want to scutter along them. I
suppose you'd attempt a GET, and if you got a 404 give up. I can't remember
the sequence for content negotiation, but you'd be hoping to get something
with the mime type "application/rdf+xml". If you got anything else, you
could maybe add a triple saying that URI was a Document. If you got a URI
then you could add those statements to your store - there might be some with
rdf:about="", which could be particularly useful.

> JG> - the TAG may have more of an answer about this.

May have, errm, eventually - it's definitely in their scope, and they would
be the best authority, the nearest I can remember is [1] (but I can only
take their material in small doses). They have got quite a pile of
relatively fundamental (difficult!) stuff on their plate.

> So, what people (and agents) would like to know, is wheter to resolve
> that uri or not and what to expect. Ok. And the person or tool should
> know that either asking something as clever as DNS but for rdf schemas
> (we don't have that yet, right?) or ask nobody.
>
> So the information should at best be in the document itself or in a
> linked schema.
>
> So either we add some kind of statement like this
>
> N3:
> @prefix rdfe http://rdf-extension#
>
> <my special uri> a rdfe:resolveMe
>
> or the rdf community makes the definitive decision that rdfs:seeAlso
> (or rdfe:resolvable) is *the* definitive property name for this
> purpose. Other properties are then required to subclass that magic
> property. So, in order to get along: Who decides such things?

I'm not sure, but I suspect that approach runs against the orthogonality
principle [2]. Any valid HTTP should resolve, but what you get might be a
404. Maybe.

In the case of schemas, correct me if I'm wrong, but the 'approved' approach
is to provide a human-readable doc (possibly with linked/embedded RDF
equivalent) if a GET is done on the namespace URI.

Cheers,
Danny.

[1] http://www.w3.org/2001/tag/doc/whenToUseGet.html#uris
[2] http://www.w3.org/TR/webarch/#orthogonal-specs




More information about the foaf-dev mailing list