[Bug 302679] New: [kmail] composer: better language identification

[Marek Otahal]

https://bugs.kde.org/show_bug.cgi?id=302679

            Bug ID: 302679
          Severity: wishlist
           Version: 4.9 pre
          Priority: NOR
          Assignee: kdepim-bugs at kde.org
           Summary: [kmail] composer: better language identification
    Classification: Unclassified
                OS: Linux
          Reporter: markotahal at gmail.com
          Hardware: Other
            Status: UNCONFIRMED
         Component: composer
           Product: kmail2

For spellchecker to work, proper language must be set. It would be nice if this
could be done seamlessly without user intervention - language identification. 

Two techniques come on my head: 
A/ determine by keyboard layout used
    + easy to code
    + instant
     - some languages you can type in latin (en keyb)

B/ some AI technique for language identification
    there should be some tools available, some quick solution is: for each
supported lang, make a list of top 200 used words. For the new text you
type-pair it with list(lang) with most common words. 
    I could code that easily, I just dont know how to integrate that in kmail.
    +does not suffer the "latin issue": eg. Czech - words 
dobrý  (good)
český  (czech)
člověk (person)     are 100% correct

but usually can be written as dobry cesky clovek - which could be corrected by
spellcheck but is for sure czech lang. and people use it in emails. 



USECASE: 
mostly would benefit multilingual environments or workspaces where your native
language is other than English but corporate communication is in Eng. 

REALISATION: 
I suggested ways to language identification above, for GUI part I would like a
checkbox 'automatic language identification' in the Spellchecker menu. maybe a
choice between the keyboard and AI version. 

DIFFICULTY: 
should be easy :)




Reproducible: Always

-- 
You are receiving this mail because:
You are the assignee for the bug.