Re: SOUNDEX searches
Posted in 1995
> mkuhn@rhlab.UUCP (Michael Kuhn) writes: > > The Soundex routine has been discussed and posted and archived. This works to some extent for American names. It was developed at the beginning of this century for use in paper based archives! It works best for old names (from the era when it was developed). It is not very usefull in most countries outside the US and possibly other english speaking countries. > There was a different routine mentioned in the past called Metaphone. > This algorthim was developed by Lawrence Philips and described in > "Computer Language" Dec. 1990 issue. > > I have the issue. I have converted the Pick BASIC routine to C. I > am currently evaluating this algorthim as compared to Soundex. > I will eventually get the code in the Informix archive. We have done the same for a Norwegian version. The Metaphone algorithm is based on pronounciation. That means there is no way you can make one internationally usable version. Even for english names you will have to have a special version for each language. Idealy you should also have special versions for each dialect, but it will give reasonable results without. > However, there are some discrepancies between the transformations in > the article and the BASIC routine. > > Lawrence Philips no longer works at location indicated in article. > > In the next day or so I will have the BASIC code and text from the > article typed in. > > Here's the deal. > > If anyone is interested in playing with the C routine and discussing some > of the issues involving phonetic searches I will be happy to > E-mail the routine. It can be called from 4GL as that is how I am > currently testing it. > > I would like to speed up the analysis of inconsistences in the > published article and get my programs working a little better than > the Soundex produces. > > If you have an interest please reply. > > to: > mkuhn@rhlab.com The development of metaphone for each language requires such an extencive reasearch that we don't want to give away our routines. We might be willing to sell it though. We also soon expect to have a Swedish version. We will also be happy to test your routine and discuss it with you as we would like to have an english version as well. Nils.Myklebust@ccmail.telemax.no NM-data, Dalsbergstien 7, N-0170 Oslo, Norway My opinions are those of my company