sul...@yahoo.com wrote:
> Hi, ich arbeite sehr gern mit dem Addon für Firefox 3.0. Nun möchte
> ich fragen, ob es möglich ist, eine aktuellere Version vom HanDeDict-
> Datenbestand für Moji zu erhalten/zu generieren. Die gegenwärtige
> Versionsnummer lautet
>
> Moji-SCnDe 0.6.20071013
>
> und ich schließe daraus, dass es eine Version vom Oktober letzten
> Jahres ist. Das HanDeDict ist ein ziemlich dynamisches Lexikon und
> eine aktualisierte Version wäre daher inzwischen schon wieder sehr
> wertvoll.
I have downloaded the current version of handedict and generated new
moji dict files. They are on the mozdev server now. You should be able
to get the updated files through the normal firefox add-ons update
check. I can't tell how mozdev handles the mirrors now. I may be
possible that you have to wait a while until you'll find a mirror from
which you can download the updated files. If it does not work
immediately, try again in a couple of hours.
You can also try the direct download from moji.mozdev.org although I
think you'll have the same mirror problem there.
> Außerdem möchte ich fragen, ob auch eine Abfrage Deutsch-Chinesisch
> möglich ist, ich kriege es momentan nur einseitig Chinesisch-Deutsch
> gebacken. Oder habe ich da jetzt das ganze Prinzip grundfalsch
> verstanden.
(Question related to reverse lookups german to chinese)
Generally, moji is able to support any search in any direction as long
as we have a edict for it. And that is already the main problem: There
are not edict files available for this direction.
Moreover, the edict file format does not really support it as it uses
spaces to separate the different parts for each dictionary entry. The
search word is the first entry. This works fine for Japanese and
Chinese, even if the search word contains a space as long as it is
encoded as full-width space and not a standard latin character set space.
An automated reverse search in existing edict files is very difficult.
Firefox uses JavaScript which is not extremely fast. edict is a very
simple plain text format and no database. Arbitrary searches through
arbitrary fields are therefore very inefficient on edict.
In addition, automated reverse search or reverse dictionary generation
results obviously in a very bad dictionary quality. Many dictionary
entries do not only contain simple, single word translations but also
some more extensive explanations. Generating meaningful reverse edicts
is difficult. I don't know exactly how it is done in general but AFAIK
any dictionary in any direction requires a lot of editing.
Cheers,
Gerald