EARL use scenario / technique

Hi, I am not sure where this goes in the techniques document exactly, but I
thought I would outline some possiblities first:

1. Testing for accessibility:

An author runs the "test accessibility" function on some content.

There are some automatic checks done, to find things like different images
with the same non-null alt-text (suspicious?), and a lack of any structure
element beyond except paragraphs, suggesting a lack of proper structure use.

The author is asked to confirm the text equivalents of each image as being
sensible.

The Tool also does an automatic test for reading level, but has no function
for doing anything more. It records its own result.

Each result is recorded as an EARL assertion.

The author then edits the content, changing some of the text and images, and
runs the test again.

There are confirmations now for 90% of the text alternatives, recorded in
EARL. So the author is only asked to confirm the remaining 10%.

The reading level comes out as higher than before, and the author is notified
of this. The tool also annotates the existing report of reading level with
one that says the Tool is suspicious thatthe requirement has not been met.

The author now opens the page in a different tool, which can help with some
text editing. This tool runs an automatic process examining the styling
properties, and suggests a potential structure model (based on size and
weight of fonts, and positioning). The author confirms that this model is
good, and the existing report of suspicious lack of structure is annotated
with the updated report that the user has confirmed the structure model.

The tool also runs a paragraph by paragraph check of the text. In each case
it uses dictionary lookup and thesaurus substitution to suggest a replacement
with a lower reading level requirement. The author agrees or disagrees at
each paragraph level, and it is noted that the author has now made the
judgement on a paragraph by paragraph basis.

What are the specific techniques here?

record the reading level of content at each check. If the level increases, a
tool may record a suspicion that the relevant checkpoint is not met. If the
author is asked explicitly, it may record that the author claims the
checkpoint is met. For abetter implementation, only do this after providing
some repair attempt.

record author's satisfaction with text equivalents. Don't prompt the author
to confirm these again if there is a recorded answer and the content has not
changed.

chaals






-- 
Charles McCathieNevile    http://www.w3.org/People/Charles  phone: +61 409 134 136
W3C Web Accessibility Initiative     http://www.w3.org/WAI    fax: +1 617 258 5999
Location: 21 Mitchell street FOOTSCRAY Vic 3011, Australia
(or W3C INRIA, Route des Lucioles, BP 93, 06902 Sophia Antipolis Cedex, France)

Received on Monday, 22 October 2001 10:56:44 UTC