[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Arabic spell-checker



Salam,

I've been working on an Arabic spell-checker for awhile now. I will be basing
the lexicon on your morphological analyzer. However, I would like to ask your
opinion on what I'm doing.

Duali (http://www.arabeyes.org/project.php?proj=duali) will do the following
when it parses a file to spell-check:

1. Strip prefix and suffix
2. Verify stem is a valid root
3. Match root against what's in dictionary (dictionary would only include root
   words, along with fixed and arabized terms)
4. If no match, stick prefix and suffix back on, and match against dictionary
   of fixed/arabized terms
5. if no match then suggest, etc (mark as bad spelling).


I am 70% through the implementation (I thought it was higher than that, but I
keep finding new things I missed :(..). 

Do you see this as an appropriate approach? How do you think it is best to
deal with bad pref+stem+suf ? ie. if the combination itself is wrong, Duali
would have very little input on what would be the correct spelling. Do you
have any suggestions?

Thanks
-- 
-------------------------------------------------------
| Mohammed Elzubeir    | Visit us at:                 |
|                      |  http://www.arabeyes.org/    |
| Arabeyes Project     | Homepage:                    |
| Unix the 'right' way |  http://fakkir.net/~elzubeir/|
-------------------------------------------------------