Hi all, Does anyone have any experience using the Double Metaphone algorithm? If so, how does it compare to Soundex? Is it worth setting up all the functions required to use it, or should i just stick with soundex? (If you want to see more on Double Metaphone, check out: http://www.winnetmag.com/Article/ArticleID/26094/26094.html Thanks, Ben
Review this articlehttp://www.codeproject.com/string/dmetaphone6.asp found in my collection. Satya SKJ Moderator http://www.SQL-Server-Performance.Com/forum This posting is provided “AS IS†with no rights for the sake of knowledge sharing.
Thanks for the link...I have read a bit more on double metaphone, and have found a lot on calculating the keys, but not anything on comparing them. Is there a specific method i should use for comparing the keys and assigning some kind of similarity rating?
Also, does anyone know how the 'Difference' function calculates similarity? DOes it just compare how many of the characters in the soundex strings have the same character in the same place?
Ben, I'm not sure about the difference function however there is an algorithm called "simil" which can be used to calculate the similarity of two strings. More info here Best regards, Gabe www.identacheck.com
What kind of match and margin of error do you allow? Double Metaphone is good when you need better quality than soundex but it will cost performance (not good in quick search). The draw back with soundex is the first character or the name/word is foreign (non-english). if you find a way to address these common issues in soundex, i guess you're fine (except for high-quality match). If you really need higher matching rate, then, the technology will be similar to master client index formulation which uses other methods and formulas. The Empowered Information Systems Claim Analyzer (www.empoweredinformationsystems.com) uses mathematical and statistical formulas aside from the AI technology. Of course, that the business requirement when it was designed was very high. But for searches, use a modified soundex to handle the short comings I mentioned. May the Almighty God bless us all! www.empoweredinformation.com