On 15 October 2010 15:22, jschmied <nabble(a)juergenschmied.de> wrote:
Hi!
I have way better results with DoubleMetaphone. It's available in apache
commons codec.
Indeed, although testing for 100 different contexts of the letter 'C' alone
sounds quite heavy.
You need a two step approach:
First select the canidates with phonetic keys (Soundex/DoubleMetaphone)
Then filter all canidates with the original search value with something
like
JaroWinkler and a defined threshold.
But the "soundslike" operator in Drools is a binary relational operator. I
cannot relate your two-step approach to that. Drools currently has no
intent of exposing the word-to-key algorithm. But who knows - if there's
good cause...
-W
juergen
--
View this message in context:
http://drools-java-rules-engine.46999.n3.nabble.com/soundslike-report-on-...
Sent from the Drools - Dev mailing list archive at
Nabble.com.
_______________________________________________
rules-dev mailing list
rules-dev(a)lists.jboss.org
https://lists.jboss.org/mailman/listinfo/rules-dev