[Top][All Lists]

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

[aspell-devel] Re: regarding autocomplete for hindi words

From: Jose Da Silva
Subject: [aspell-devel] Re: regarding autocomplete for hindi words
Date: Thu, 18 May 2006 20:55:19 -0700
User-agent: KMail/1.7.2

On May 18, 2006 02:18 am, you wrote:
> Hello,
>       I am trying to develop an autocomplete mechanism
> for hindi language.

If you can make it work for hindi, it might be worth making it work for 
other languages :-)
Likewise, if it already exists for another language, it would be even 
better if it works for hindi as well.
If you look at some of the more popular desktops like Gnome, or KDE, 
they are very nice to use since they can run different languages at the 
same time... you might want to get involved in adding hindi where you 
can to them.

> suppose we have a set of hindi 
> words in a database,then how do we retrieve the
> probable set of  words for an incomplete word?

You make the wordlist so that the more popular words are 1st instead of 
making it absolutely alphabetic. For example: "and" is used more often 
than "add", so "and" is in the list before "add" if we are talking 
about English.
However, you still need to make your wordlist alphabetic if you are 
going to make it run fast, so if I typed "an", I already know I was not 
looking for "add"
So, you need to make it a combination of the two.
Look carefully at the wordlist and the popularity of each word.

Take a look at how the aspell wordlists are organized. Take a look at 
the aspell website, it points to several dictionaries and resources.

> Is it enough to match the unicode characters of the
> typed
> word and match the unicode sequence with the unicode
> sequence of the incomplete word, or are there any more
> complications?

There are always complications :-) but you need to begin somewhere.

> Bye,
> Ravi Sastry

What you ask might already exist for another language.
If it exists, it would make sense to make it better to work for that 
language and hindi as well.
Maybe someone on the list has a suggestion.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]