Re: [Indlinux-group] how fit is hunspell for your language's spell checking?

[Guntupalli K. <kar...@in...>]

On Wed, 03 Oct 2012 13:32:32 +0530
ஆமாச்சு wrote:

> On Wednesday 03 October 2012 08:35 AM, Parag N(पराग़) wrote:
> >
> > Can you explain this what exactly you are trying to do?
> 
> here I had expressed my opinion that, hunspell won't be sufficient 
> enough to take care of Tamil spell checking requirements.
> 
> I want to know, opinions of other language people here.
> 

As such what we have been using till now is merely word list with
suffix rules. Not much research is gone into spellcheckers beyond it
(atleast in public domain).

> 
> > Do you have any README or information about your project in
> > English somewhere? :) 
> 
> Will add as I progress, there.
> 
> > The last updated Tamil hunspell tarball you gave to me contains
> > 112425 words. Do you find those words are not good or need
> > proofreading?
> 
> Well, I too contributed to that list. But seldom words exist like
> that in reality. I am aware of the rules for affixes-suffixes to
> the words.
> 
> What I plan to have is a word list that is classified primarily
> into nouns & verbs. I feel that will help taking much closer to the
> requirement.
> 

This would be good thing to have, word list now is a flat one with no
differentiation/classification 

> Spell checking is beyond correct word correction.
> 
> gender wise sentence construction, how sentences vary with
> salutations like mein, thoo, aap etc.,
> 
> can hunspell actually take us to these levels?
> 

What you mean to add is like grammar check.. if you could generalize
it while doing it for tamil, would help too.

Karunakar