W3C home > Mailing lists > Public > www-rdf-interest@w3.org > September 2002

Re: parsers that don't need rdf:RDF?

From: Nikita Ogievetsky <nogievet@cogx.com>
Date: Fri, 27 Sep 2002 23:18:35 -0700
Message-ID: <017001c266b6$e0faa980$1101a8c0@VAIOCOGI>
To: "Dan Brickley" <danbri@w3.org>
Cc: "Jeremy Carroll" <jjc@hpl.hp.com>, "Bob DuCharme" <bobdc@snee.com>, <www-rdf-interest@w3.org>

Dan,

Dan Brickley  wrote:
>
> On Fri, 27 Sep 2002, Nikita Ogievetsky wrote:
>
> > Indeed, this "rdf:RDF" requirement makes otherwise compact RDF
annotations
> > somewhat bulky.
> > I would propose that processors may derive RDF-ness of the metadata
> > by resolving namespace URI to see if it is an RDF schema.
> > Otherwise it may just ignores the metadata.
> >
> > Currently most RDF processors require "resolvability" of  schema URI-s.
> > With this proposal they will only assume that they are parsing an RDF
> > fragment
> > if the namespace resolves to an RDF schema.
> >
> > --Nikita.
>
> One of the design constraints on RDF was that the content (as
> graph) of these documents should be self standing, ie. not need you to go
> retrieve, parse, interpret further documents on the Web before you can
> figure out what its content amounts to. This concern also explains RDF's
> apparent verbosity: it makes everything (sometimes painfully) explicit,
> instead of relying on possibly remote external documents.

Yes, you are right. However, some RDF applications do require resolvability.
(especially those related to inferencing)
For example, DAML validator.
And it also allows caching for bad days :-)

> That said, a lot of people are interested in exploring the use of XML
> Schema annotations to map from more colloquial XML into RDF graphs.  IMHO
> there's a major role for this approach too, so long as we have at least
> one syntax for RDF that takes the self-standing view...

Actually I think that these approaches do not contradict each other.
1)"rdf:RDF" declares that all its contents is RDF.
2) otherwise, RDF-aware processor may determine RDF-ness of
the metadata by trying to either resolve its namespace URI, or by "knowing"
this namespace - for example RDF spiders may just know what Dublin Core's
namespace.

BTW, it is interesting to note that case 1) is an instance of 2): RDF is
just "known"
by RDF processors. Why not allow "knowing" of DC, RSS, PRISM, etc.?

Also, resolving schema is worth a try if some inferencing needs to be done.


--Nikita



Nikita Ogievetsky,  Consultant.     Cogitech Inc.
email: nogievet@cogx.com   phone: (917) 406-8734
web: http://www.cogx.com   Cogito Ergo XML

"Building, Aggregating and Navigating Information Systems."
New York Associated Topics Seminar. November 18, 2002
http://www.cogx.com/associatedtopics/
Received on Saturday, 28 September 2002 02:18:52 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:51:56 GMT