WCAG

For the medium-term development of Site Valet, I'm looking to
incorporate a parser based on WCAG.  As with other new developments,
it will generate XML and thence multi-format output via XSLT[1].
I'm trying to think this through as much as possible now, not
least to determine whether it will generate any relevant comments
on sbp's EARL draft.

This raises questions over which WCAG to start from, and will
doubtless raise followup issues.  I'm wondering whether WAI/ER
(which is the only WAI group I subscribe to as of now) is the
best place  to ask such questions, or whether I should contemplate
crossposting to anywhere else.

My first thought is a message structure

* Document , and location within doc (e.g. line and column,
	as in validator)
* Code element in question (e.g. <img src="bullet.gif" alt="bullet">)
* WCAG checkpoint referenced
* Severity (WCAG priority)
* Belief: is this definitely an error, probably an error, or merely
	something that's probably OK but the machine has to report?

I'd like to be able to represent belief numerically, e.g. as a
percentage value.  This would give us a full infrastructure for
combining reports from multiple sources in an evidential
reasoning (e.g. Bayesian or Dempster-Shafer) framework.

[1] The aim is that all Valet reports should be available as
    plain text, HTML, XML and EARL, as a user option.

-- 
Nick Kew

Site Valet - the essential service for anyone with a website.
<URL:http://valet.webthing.com/>

Received on Monday, 29 October 2001 18:54:10 UTC