[namespaceDocument-8] RDF and RDDL

At last week's TAG telcon I took an action to explore the use of an RDF
based approach for embedding machine readable information on RDDL documents
instead of Xlink. The comments raising concerns over the embedding of RDF
embedded in HTML raised in [1,2] I thought I expose where I've got to in the
last week. I'd also like to thank Tim Bray and Jonathan Borden for their
input and encouragement.

Quoting from the RDDL spec (http://www.rddl.org):
"This document describes the Resource Directory Description Language (RDDL).
A RDDL document, called a Resource Directory, provides a package of
information about some target, including: 

 - Human-readable descriptive material about the target. 
 - A directory of individual resources related to the 
   target, each directory entry containing descriptive 
   material and linked to the resource in question. 

The targets which RDDL was designed to describe are XML Namespaces."

Basically RDDL provides an HTML human readable descrition of each resource
related to the target, and adds a machine readable Purpose and Nature.
Purpose is encoded in an xlink:arcrole role while Nature is encoded in an
xlink:role attribute.

I have explored two main approaches: 

One is to follow the pattern of Ron Daniels Xlink to RDF Note [3]
(rddl-rdfA.html and rddl-rdfC.html) which maps xlink:arcrole (purpose) into
the predicate of an RDF statement and maps xlink:role (nature) if present
into an rdf:type property of the related resource (xlink:href and object of
an RDF statement).

The other approach is to more directly model purpose and nature as RDF
properties (rddl-rdfB.html and rddl-rdfD.html). 

In addition, I have also tried embedding the HTML fragment that describes a
related resource in the RDF as a rddl:Description property - in a way that
still renders the HTML when viewed in a browser (I'm using IE5). This
captures the HTML fragment as a literal value when processed by an RDF
processor (Jeremy Carroll's ARP via the W3C RDF Validation service
http://www.w3.org/RDF/Validator/).

N3'ish descriptions of these four variants follow:


A: (based on Xlink2RDF)
<rddlSourceURIRef> <purposeUriRef> [<relatedResourceURIRef> rdf:type
<natureURIRef> ].

B:
<rddlSourceURIRef> <rddl:Purpose> <purposeURIRef>;
                   <rddl:Nature>  <natureURIRef>;
                   <rddl:RelatedResource> <relatedResourceURIRef> .

C:
<rddlSourceURIRef> <purposeUriRef> [<relatedResourceURIRef> rdf:type
<natureURIRef> ] ;
                   <rddl:Description> "Literal HTML Description" .

D:
<rddlSourceURIRef> <rddl:Purpose> <purposeURIRef>;
                   <rddl:Nature>  <natureURIRef>;
                   <rddl:RelatedResource> <relatedResourceURIRef> ;
                   <rddl:Description> "Literal HTML Description" .

Attached for each approach is a small extract from the RDDL specification
[4] which retains the original RDDL description (for comparision) and one of
the RDF variants. I've also attached a separate file for each variant
containing just the RDF and a .PNG file of the RDF graph that results when
passed through the RDF validation service (couldn't produce working SVG's
for the variants that contained the embedded HTML - also CWM had a digesting
those as well).

The status of all this is very experimental and should not be regarded in
any sense as definitive. Something of a "what-if".

Thoughts and comments welcome, and once again thanks to Tim and Jonathan for
their time during the week.

Best regards

Stuart Williams

[1] http://lists.w3.org/Archives/Public/www-tag/2002Apr/0051.html
[2] http://lists.w3.org/Archives/Public/www-tag/2002Apr/0062.html
[3] http://www.w3.org/TR/xlink2rdf/
[4] http://www.rddl.org