Re: Adding gloss translations [Was: Re: [edict-jmdict] JMdict editing form]

["Alexandre Courbot" <gnurou@*********>]

Hello everyone, sorry that it took so long to get back to you. I have
tried to contact the maintainer au dictionnaire-japonais.com on their
forums, without success until now. I have tried to directly contact
him today.

> Sounds like a good approach to me, but "an API" is
> pretty vague, yes? It would be interesting to think
> about something more concrete. Like an api to what
> language/technology? Any cgi script that takes
> parameters is in some sense an api, but I presume
> Alexandre had something more specific in mind.

Sorry for not having been more precise. The idea behind the API stuff
is precisely not to be bound to any particular language, by using a
standard that is already largely used across the web (e.g. SOAP).
Indeed, behind it there could be a classical CGI.

It is indeed easy to use the current form to submit new entries - but
as far as I know it cannot be used to fetch existing information in
order to allow the user to easily fix an existing entry. Translators
would need to have a complete view of the different senses and glosses
in English in order to translate them efficiently. The idea behind my
proposal is to allow the client to fetch the current record of an
entry in XML form and to submit an update in the same format (or to
submit a new entry in that format, too). This way would be much closer
from the actual format of the database which would prevent ambiguities
in editing entries and also facilitate their integration (but here, I
only speak based on my own assumptions about the current submission
system). It could be used to generate edit forms (querying the DB from
the server, using PHP or any other CGI) or to easily batch-edit
entries using a program written in any language (Python comes to mind
here). With such an API, it would be easy to do tasks like including
examples sentences, associating entries that are related in some way
(自動詞 and 他動詞 come to mind), or adding JLPT levels from a list (which I
have up to level 2, in case there is interest for that). And, I also
believe it would remove some of the burden from Jim's shoulders by
allowing delegation of work. As the JMdict grows, the maintainance
work also grows and I have the feeling that the current way could be
improved.

Of course, such editing power should be limited to trusted parties,
hence the idea of using authentication (of course, review before
actual integration is also possible). If the idea is still not clear
enough, I can try to set up a small example platform. And if the idea
is accepted and help is needed for its implementation, I'm also ready
to help.

Alex.