I enjoyed Michael's mail about multi-level indexing. It's something
we've put some thought into here at NZETC, as well, since we have plans
to extend our markup in this direction. Our plan is to use the index
element, but only in a simple way, i.e. with only @level1.
I agree that using multi-level index markup would in general effectively
presuppose a pre-built taxonomy, so you wouldn't gain anything by then
devolving bits of that taxonomy back into the body of your text. Better
IMHO to build an index with simple index elements (i.e. with each
indexed item pointing to a single specific concept), and to impose a
hierarchy on those concepts independently of those indexed elements
(perhaps as an eTree http://www.tei-c.org/P4X/GD.html#GDAT or perhaps in
some external system ... a database, topic map, thesaurus, or whatever).
In this respect, it seems to me that it would be nice if <index>
elements could use pointers (rather than textual strings) to identify
the index headings or concepts. But perhaps this is more properly a case
for <rs> or <xref> or something?
I would be very interested to hear of any examples of the use of
<index>, the generation of indexes, and especially linking to external