|
|||
![]() |
|||||||||||||||
![]() |
|||||||||||||||
|
Phonetic algorithms take single words of ordinary languages like English and German, and generate keys which represent the phonemes of these words. The goal is to generate equal keys for similar words and different keys for unsimilar words, and therefore make similar words comparable for equality. As an example take the words "Phonetix" and "Fonaetic". Both aren't successfully comparable belonging textual equality. A phonetic algorithm, like Metaphone, may generate the same key (here: "FNTK") for both words, therefore making them successfully comparable belonging equality of their phonemes. Each phonetic algorithm generates algorithm-specific keys. Some algorithms use special knowledge about the phonemes used in particular languages like English. Some algorithms are more sophisticated than others. Soundex is an old and simple algorithm in widespread use. DoubleMetaphone is rather new and very sophisticated, but less well-known. Metaphone is somewhere in between. There is no "best" algorithm, because this depends on the application. Soundex is fast, but somewhat crude. DoubleMetaphone differentiates phonemes finer, but is therefore slower in execution. You have to test each algorithm with application's data and compare the results to choose the algorithm which fits your data best. Yes, of course. Phonetix is licensed under the GNU Lesser General Public License and therefore usable as-is in commercial software. Yes. You may get free email support in case of problems or suggestions. |
||||||||||||||
![]() |
|||||||||||||||
|
©2003, tangentum.com
All rights reserved.
|
|||||||||||||||
![]() |
|||||||||||||||