Orphaned annotations

I've done a little more work on a client for Annotea, with a view to
supporting annotations in Page Valet.

Since Page Valet generates a normalised representation of page markup
showing validation errors and accessibility warnings as and when they
arise in the markup, it makes sense to use annotea's pseudo-xpointers[1]
(which valet already uses) to reference annotations to the markup.

This is essentially equivalent to what any other annotea client can do,
but because Valet is a diagnostic tool, the view and its purpose differ.
As soon as I had a working prototype, it became abundantly clear that
there is a deep and fundamental flaw in Annotea: we construct long
and detailed pseudo-xpointers, but these become totally useless as
soon as a page is updated.  And annotea has no mechanism for dealing
with this, nor indeed even to detect that a page has changed.

It seems to me that annotations need some expiry mechanism for
when the annotated resource changes.  This should be qualified
by the significance of a change: for example, a page bearing todays
date may not be regarded as changed when the date becomes tomorrow.

The problem of measuring when a change is significant is one that
has been discussed in ER, where we have considered document hashing.
My prototype implementation[2] hashing on ESIS by filtering nsgmls
output appears to do the job, and might be worth considering for
Annotea:
  * Where a document's Elements have changed (as measured by
    the hash), the xpointer is invalidated.
  * As changes are made, the quality of the annotation is reduced as the
    significance of a change increases.  Quality exceeding some threshold
    could be a database search criterion; dropping below some threshold
    could be used to delete or archive the annotation.  For example,
    if the Headings have changed, there is a strong chance the
    earlier annotations are no longer applicable.


[1] They are xpointers only when a page is well-formed XML, which is
    neither usual on the Web nor required by Annotea.
[2] http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2001Dec/0029.html
    http://lists.w3.org/Archives/Public/w3c-wai-er-ig/2002Jan/0019.html

-- 
Nick Kew

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

Received on Sunday, 17 March 2002 20:37:51 UTC