Transliteral Application

for local language

[kadnan]

Google recently released transliteral application for Indian languages which convert english into local Indian languages(http://www.google.com/transliterate/indic/).

I want to make similar for Urdu language(Arabic like) which is also a Unicode language. I am just confused how to implement it. Whether I use php's builtin functions to find similar words or maintain my own database. For instance in Urdu, the term "Tum" is used for "You". Now one could write it as

1)Tum[right one]
2)Tom
3)Toum.

Offcourse I have urdu keyboard script written in Javascript. what I want that if someone writes "Tum" and hit space, php scripts generated similar sound words itself[possible]? and then my script map each character with corresponding urdu character.

I desperately need your help. Thanks

[Habtom]

I am not very sure how this is going to go, but I have once used SOUNDEX to get similar sounding words.

SELECT SOUNDEX ('Smith'), SOUNDEX ('Smythe');

If Soundex is not for mysql, you might try find the equivalent one.

Hope this gives you an idea.

Habtom

[kadnan]

which SqlDb did you use? did you stored words in db? What I want that i input a word(in English) and then it returns the words which are near to that word as I gave the example of "Tum".

[Habtom]

which SqlDb did you use? did you stored words in db?

Yes it was stored in a database. I can barely remember this but I think it was Oracle. You browse around for Mysql, probably that works as well.

Read about it, and let me know what you come up with.

[kadnan]

db usage wouldbe very expensive. Imagine several users are making hundreds of requests per draft.

[Habtom]

db usage wouldbe very expensive. Imagine several users are making hundreds of requests per draft.

How did you know? Did you make a test or it is just an assumption?