Lists Home |
Date Index |
I'm pleased to announce that the W3C RDF Core Working Group has
published a new version of the RDF/XML working draft revising the
syntax originally specified in the RDF Model and Syntax W3C
RDF/XML Syntax Specification (Revised), edited by Dave Beckett
W3C Working Draft 8 November 2002 (published 2002-11-12)
This syntax re-specifies the original grammar in terms of XML, XML
Namespaces, the XML Information Set including support for XML Base.
The main changes from the previous 25 March 2002 draft are as
follows. I've added some comments for background, but there are lots
of links in the document to the issues and discussion.
* Section 2 "An XML syntax for RDF" expanded with many more examples.
This is an attempt to explain the RDF/XML syntax including the many
(too many) forms of abbreviation in RDF/XML with the use of examples.
We don't suggest you use all of them.
We have had the suggestion that we describe a simple profile that you
can stick to if you don't care. This might end up similar to the
approach Adobe did with their use of RDF/XML in XMP, which has been
described as RDF/XML-7 - they don't use 7 forms of the abbreviation
We particularly ask for feedback on this section since it has a lot
of new words and examples, We need to know if this section helps.
The feedback address is at the top the document, where the archived
comments can be read too.
* RDF typed literals added using a new rdf:datatype attribute.
So now you can mark literals as being of type XSD integer, for
example. This new term is explained more in the RDF Concepts and
Abstract Syntax working draft at
* rdf:nodeID attribute added to allow referencing of blank nodes.
This makes going between RDF graphs to/from RDF/XML a lot easier.
The main caveat is property URI-references that can't be written as
XML Qnames. More detail is given on this at
* rdf:parseType="Collection" added for collections of nodes.
This is modeled after daml:Collection in the DAML+OIL syntax (based
on RDF/XML), generating a lisp-like triple structure of rdf:first,
rdf:rest, rdf:nil etc.
The Web Ontology Working Group and others asked for this. It
hopefully will makes things easier for OWL and in general,
inferencing over or describing such structures in triples.
* Some non-namespaced attributes were returned to the grammar:
ID, bagID, about, resource, parseType or type
This makes a lot of existing RDF/XML work without warnings but apart
from them, in particular we wanted to ensure that the Adobe XMP
profile of RDF/XML and that used in Mozilla remained valid.
The detailed changes to the document are recorded at
These syntax changes are already being implemented in parsers for
several languages (so you probably won't have to), and I'm expecting
there will be several (3-4) 90%+ done quite soon. To help with this
there is a set of test cases with N-Triples test format described in
and the tests themselves at
that we will be using to help with this.