At 09:04 PM 9/7/01 -0700, Mark Davis wrote:
>I disagree. What you want is a merged database field. See
>http://www.macchiato.com/slides/icu_collation.ppt
>
>Mark
Mark,
David took the remainder of our discussion off the alias. I won't repeat it
here, just to note that we've agreed that merged database fields are the
answer to (some) of the scenarios that we've discussed, but that there are
cases (like indexing a mixed corpus where both naive and naïve occur) where
it might indeed make sense to ignore accent differences altogether -
although, as is often the case, dictionary-based pre- or post processing or
manual adjustments might give better results yet.
Thanks for your pointer to the presentation.
A./
This archive was generated by hypermail 2.1.2 : Sat Sep 08 2001 - 15:56:22 EDT