Re: Soundex in other language

  • From: "Dennis Williams" <oracledba.williams@xxxxxxxxx>
  • To: LegeDoos <legedoos@xxxxxxxxx>
  • Date: Tue, 29 Apr 2008 08:53:30 -0500

Rob,

You can find the algorithm at: http://en.wikipedia.org/wiki/Soundex
My suggestion would be to experiment with it yourself, maybe first in
English to confirm your understanding, then in Dutch. My understanding is
that English and Dutch share common roots so Soundex might do reasonably
well as-is. I just saw a TV program on the origins of English and
they explained Frisian was the closest existing relative to English.

Many people misunderstand what Soundex is. The only goal is to take the same
words with variants of spelling (and misspelling) and rank them in
closeness. Previously I worked for a mail-order company that would give a
small amount of credit to nearly any new customer. Obviously some existing
customers (who was probably behind on payments) would call up and try to
pose as a new customer. Normally they would intentionally misspell their
name, knowing that the postal employees would overlook small misspelling.
Normally a small misspelling of a name will defeat a computer search. With
Soundex function-based indexes, the system could usually find whether this
was actually an existing customer.

Dennis

Other related posts: