W3C home > Mailing lists > Public > www-tag@w3.org > September 2003

Re: New RDDL draft

From: Jonathan Borden <jonathan@openhealth.org>
Date: Wed, 17 Sep 2003 13:29:29 -0400
Message-ID: <3F6899F9.8030906@openhealth.org>
To: Tim Bray <tbray@textuality.com>
Cc: WWW-Tag <www-tag@w3.org>

Tim Bray wrote:

> My action item is half-done, I need to do the following:
>
> - add a "hello world" example near the top of the doc
> - produce a normative transformation into RDF, probably XSLT

There are a few nitty gritty issues which raise their heads when 
considering the RDF transformation.

First I will outline what should be a straightforward protocol for the 
transformation, then I will point out where these issues fail to get 
properly addressed:

Assume a RDDL fragment (from Tim's most recent draft):

<a rddl:nature="http://www.rddl.org/"
   rddl:purpose="http://www.rddl.org/purposes#directory"
   href="http://www.rddl.org/natures">http://www.rddl.org/natures</a>

1. The subject of an RDF statement is derived in the following fashion:

2. If the there is an _id_ attribute e.g. <a id="foo" ...> this should 
be the fragment identifier of the RDF subject URI.
3. The URI part of the RDF subject URI should be the current base URI 
(absolute).

4. The predicate is the value of the _rddl:purpose_ attribute.
5. The object is the value of the _href_ attribute.
6. If a rddl:nature attribute is present, this generates a second triple 
of the form:
[value of href] rdf:type [value of rddl:nature]

This generally works but stumbles when a namespace name is not simply an 
absolute URI, rather contains a '#' either an ending '#' or a fragment 
identifier part.

That is to say, suppose a namespace name of the typical RDF type e.g. 
<http://example.com/rdf#>.

The problem is that a user agent will strip off the ending '#' when 
dereferencing the URIref and the base URI will simply be 
<http://example.com/rdf>. How do we tell the software performing the 
RDDL -> RDF transformation that the intended namespace is actually 
<http://example.org/rdf#> as opposed to <http://example.org/rdf> when 
the RDDL document itself does not contain this information? Do we need a 
specific attribute e.g. rddl:namespace="http://example.com/rdf#" which 
states the one and only namespace described by the specific RDDL 
document (you see the problem that each RDDL document would only be able 
to describe a single namespace in this case).

Am I missing a straightforward way to tell the software what the correct 
intended subject of the RDF statement is without embedding this in the 
document itself?

Jonathan
Received on Wednesday, 17 September 2003 13:29:40 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 26 April 2012 12:47:20 GMT