Re: soundex()

From: Lawrence James <James.Lawrence_at_epamail.epa.gov>
Date: Tue, 1 Nov 1994 20:55:03 GMT
Message-ID: <James.Lawrence.33.000FEB5D_at_epamail.epa.gov>


In article <8TYTBKVM_at_news.GUN.de> awinkler.gun.de writes:
>From: awinkler.gun.de
>Subject: soundex()
>Keywords: oracle soundex source
>Date: Sun, 30 Oct 1994 19:54:36 GMT
 

>does anyone know if there is a source for calculating the
>soundex ?

Yes, I'm going from memory here so forgive me if I'm wrong. There is a three volumn set of books about computer programing by a guy named Knutt. I think the set title is 'The Art of Computer Programming'. In one of those is the algorithm for it and I have reproduced it successfully in COBOL (long ago so don't ask for a copy)

It goes something like drop all vowels, drop all 2nd occurances of letters. Finally there are groups of letters with each group being assigned a number. The letters are ones that sound alike such as c + k, etc. I'm sure this is only an approximation. The book goes into various algorithms for doing this sort of stuff along with the math behind it. It's also really old! (Depending on your definition of 'really old')

I will look around my stuff and see if I find anything more accurate and will post if I do. If not, this should give you some leads.

Lawrence..... Received on Tue Nov 01 1994 - 21:55:03 CET

Original text of this message