EARL as DOM+assertions model?

I seem to have lost my route to openprojects.

An idea we touched on is what Chaals and I characterised as
DOM + assertions.  This is effectively how Page Valet's accessibility
checking works: it constructs a document tree, and applies tests
on Element and Attribute nodes in the tree (an approach inspired by
my experiments with Schematron back in about October).

The native XML output reflects this approach: the document tree is
represented by element and attribute nodes, and nodes are generated
within the elements by any test that flags a warning:

<element name="html">
  <attribute name="lang">en</attribute>
  <element name="head">
    ....
     <element name="something-bad">
        ...
	<test
		fuzzypointer="html[1]/head[1]/something-bad[1]"
		severity=... confidence=...
		href="something in WCAG"
	>Don't do that!</test>
     </element>
  </element>
  ....
</element>

Turning that into valid RDF while preserving the structure should
AFAICS be reasonably straightforward.

Now, if as (I think) Chaals was suggesting, EARL could take this kind
of structure, it could become the *native* format not only for Page
Valet, but for more-or-less any tool based on [HT|SG|X]ML parsing.

We already know the above with the addition of metadata (assertor,
date, etc) can be transformed to EARL-as-currently-defined by XSLT.
We also know that there is a wealth of XML tools capable of
processing the above.

Any thoughts?  What happens if I hack up something using EARL namespaces?

Oh, and what happens if we substitute Annotations for Assertions in
my argument?  Or, better, Annotations which are also EARL Assertions?

-- 
Nick Kew

Site Valet - the mark of Quality on the Web.
<URL:http://valet.webthing.com/>

Received on Wednesday, 20 February 2002 14:46:59 UTC