> I'm not sure why you need another engines. > > Basically spell engines by applications use Myspell > dictionary files. > I checked myspell dictionaries now and it seems > Turkey and Finnish dictionaries are available. > > http://ftp.services.openoffice.org/pub/OpenOffice.org/ > contrib/dictionaries/ku_TR.zip > http://ftp.services.openoffice.org/pub/OpenOffice.org/ > contrib/dictionaries/fi_FI.zip > > If you proceed the ARC process, I can include the > dictionaries in SUNWmyspell-dictionary-$LANG. > > fujiwara >
Hi i am one of the zemberek developer. we were contacted recently about this subject, but i just had time to see here. First a correction, there is no Turkish myspell dictionary available. the link you pointed is not for Turkish, it is for Kurdish (a completely different language). currently there are two open source sane Turkish spell checker available. As noted, - Zemberek, a special java library using morphological analysis algortihms. it has posibly %99 success ratio on spell checking. - a recent work we did; "tr-spell" an experimental turkish hunspell dictionary. This is basically a huge wrod list (1.2 million words) created with the help of Zemberek. may be converted to myspell too. ( http://code.google.com/p/tr-spell/ ) it may have %95-96 accuracy. The first one, Zemberek is very accurate comparing to the second one, but embedding it to gnome may be a little tricky. Thankfully there are gnome enchant patches available for several linux distributions. i will try forwarding the related people to here and people working on this subject soon. thanks for the work in open solaris. thanks. -- This message posted from opensolaris.org
