[Fedora-i18n-list] Re: Poor Quality of Japanese Conversion

[Akira TAGOH]

>>>>> On Wed, 08 Sep 2004 18:14:31 +0900 (JST),
>>>>> "AT" == Akira TAGOH <tagoh at redhat.com> wrote:

>>>>> On Tue, 07 Sep 2004 15:52:09 -0400,
>>>>> "morpheus" == morpheus <morpheus at post.harvard.edu> wrote:

morpheus> Sorry, I acciedntally sent the message incomplete, due to the poor
morpheus> quality of the user interface...ironic, isn't it?

morpheus> Hi, I know this is not strictly an IIIM issue, but can anyone recommend
morpheus> a good quality Japanese conversion engine.  Using Canna is like using a
morpheus> ワープロ from 1975.  It always suggests the most unusual and unlikely kanji
morpheus> instead of the most common ones, for example, when I enter:
morpheus> さいしん I get 斎信 instead of 最新 which is far more common.  Similarly, for 
morpheus> らいしゅう I get 来集, for こんしゅう I get 今秋, etc. etc.

AT> Well, it depends on which dictionaries are you using. I
AT> fixed similar problem before. but it still appears for some
AT> reason in IIIMF and not updated for default.canna file. I'll
AT> update Canna package to improve the kanji conversion on
AT> IIIMF then.

er, it also depends on the learning by frequency of use in
the dictionaries. although it's default behavior -- I could
reproduce it -- the dictionaries probably needs to be
optimized. however that way is not necessarily good for all
of people, but anyway.

--
Akira TAGOH