[Project-ideas] gsoc 2012 project idea---A Glossary Tool to index all available localizable strings for Bengali across various FOSS projects and, other available corpus

[Subhobrata Dey]

Hello ma'am,
          thank you for your reply.
Well,i propose a formal solution for this project based on your reply &
studying the translators like lokalize,virtaal etc.
First of all,i went through the lokalize documentation & the source
code(since i love kde) from which i mainly got the idea for this project.
While lokalize uses paragraph-by-paragraph translation technique, i think
using a sentence-by-sentence translation technique followed by using a
word-by-word translation is far more efficient.
The idea mainly came after reading this article:
link <http://Quick%20View>
Here i'm elaborating the idea:
We can classify english-bengali translation into the following ways:
1>Assertive sentence
I will go to bazaar.
আমি বাজারে যাব
Here the predicate goes to the end & the subject after it comes forward.
If I were a bird!
যদি আমি পাখি হতাম
these are same class of sentences....
2>Interrogative sentence
Are you reading a book?
তুমি কি বঽ পডছ?
Here predicate & subjects are swapped
3>Ordering
Give me the book!
বঽটা আমাকে দাও
There are a few common exceptions like, for 'the' ' টা' can be used.for
'please' 'দয়া করে' can be used.
But majority of the translations must fall into these categories.
So,my idea is for each sentence in english we first need to decide what
kind of a sentence it is.Then apply word-by-word translation on it to
convert it to bengali & take the help of glossary to substitute english
subjects & predicates with their bengali version(However a few may not have
their bengali translations,appropriate measures need to be taken for
that.Haven't thought about that as yet :P).This will obviously not give the
correct result.So,from the info regarding the kind of sentence we can swap
the bengali words to get the right result!!Thus sentence-by-sentence of a
paragraph is translated.
Apart from this,i also saw that there are shortcuts for words & phrases in
lokalize while giving input which i think should be implemented for the
sake of users.
In this way, a successful glossary cum translator can be created.
Finally,there can be mistakes & because of that i'm posting this in the
mailing list.

Btw,I'm not posting the section "about me" as of now.Please mention if you
need that or any other info from me.....i'll be extremely happy to provide
them to you.

Thanking you,
-- 
subhobrata
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ankur.org.in/pipermail/project-ideas-ankur.org.in/attachments/20120322/163c1b03/attachment-0003.htm>