And this is a key advantage of using the mapping char filter rather than the
simple ASCII folding token filter - you can easily go in and modify the
mappings for application/domain/environment-specific character mappings such
as these.
-- Jack Krupansky
-----Original Message-----
From: André Widhani
Sent: Monday, June 17, 2013 4:27 AM
To: solr-user@lucene.apache.org
Subject: AW: Best way to match umlauts
We configure both baseletter conversion (removing accents and umlauts) and
alternate spelling through the mapping file.
For baseletter conversion and mostly german content we transform all accents
that are not used in german language (like french é, è, ê etc.) to their
baseletter. We do not do do this for german umlauts, because the assumption
is that a user will know the correct spelling in his or her native language
but probably not in foreign languages.
For alternate spelling, we use the following mapping:
# * Alternate spelling
#
# Additionally, german umlauts are converted to their base form ("ä" =>
"ae"),
# and "ß" is converted to "ss". Which means both spellings can be used to
find
# either one.
#
"\u00C4" => "AE"
"\u00D6" => "OE"
"\u00DC" => "UE"
"\u00E4" => "ae"
"\u00F6" => "oe"
"\u00DF" => "ss"
"\u00FC" => "ue"
André
=
