Just a few comments on this indexing discussion...
I'm finding the <index> section of the <add> quite useful to
collocate authoritative forms of names, places, etc. together
when there were many instantiations of the name (in various
non-authoritative forms) throughout the <component> elements.
Each <indexentry> can have one or more pointers to the <components>
which used the name using ID/IDREF mechanisms. You can also
use the <controlaccess> element in a give <component> element
to list the authoritative forms of the names, if you prefer
these to be at the folder level. This, in combination with
the "sources" attribute for names, would seem to address a
lot of the concerns I've heard.
Richard's proposal to use a 'reg'-type attribute works fine if
the name only occurs once or twice, buts it's not so great when
it occurs fifty times in a finding aid. First there's the extra
keying involved, then there's the issue of needing the index to
be _in_ the finding aid for exchangability, different search
engines, etc. (PAT indexes are _outside_ the document). We need
to be able to accomplish our retrieval goals _no matter what_
search engine is at the end. I suppose we could all write perl
scripts to generate an <index> from all the <name> elements and
append it to the finding aid before loading/indexing, but that
assumes everyone has a perl programmer handy and can take apart
their workflow to add such a step. There's no way of getting
around some rekeying, I fear, but using the <index> or <control
access> mechanisms provided will minimize this extra work.
_____________________________________________________________________________
MacKenzie Smith [log in to unmask]
Harvard University Library %\%\%\%\%\%\%\%\%\%\%\%\%\%
-----------------------------------------------------------------------------
|