Re: [dict-beta] wikipedia dict file

[Piotr B. <ba...@o2...>]

[again, I sent this privately only; resending to the list, with a minor
edit]

Hello Andrew and Others,

Andrew Dougherty writes:
> While browsing legal torrent download sites, I came across Wikipedia
> as an Mdict file.  I tried to convert it but no luck.  So I just wrote
> some perl scripts to convert the Wikipedia dump to a dict file(s).
> This is taking a while.  I think I will generate 2 versions, one the
> full version and the other using automatic summarization to reduce its
> size.

Sounds nice. I know that {en,fr,de,es,it}wiki is accessible via DICT,
from aioe.org, but I have no idea how current their databases are. Does
anyone here know anything about that?

It would be good to have, let's say, a weekly dump available as .dict --
I'm thinking of the promotional aspect of this, at least. But I have no
idea how aioe.org does that and whether they roll up their own databases
or use someone else's dumps.

> Is there any interest in this?  Has this been done to death?

It hasn't been done to death, I don't recall any discussion on this in
the list archives. As for interest... I know that Aleksey concentrates
on the software, and that Rik doesn't (apparently) have the time for the
web site and the like. Maybe the ftp server at dict.org could (?) host
Andrew's wiki-dicts, if those at aioe.org are obsolete? I don't know how
this is handled and I have no idea who took (takes?) care of updating
DICT databases, and whether any such person ever existed.

> Also, I've made a slightly larger Irish dictionary than freedict's
> irish english dict using some sources on the web.  I doubt they are
> free to reuse.

So -- just making sure -- we can't hope to get this for distribution via
freedict?

Ah, on a slightly different though related note: I've been thinking of
grabbing wiktionaries for freedict, as separate language-pair packages
(they have rather amateurish lexicographic control, on the average, so
are not to be fully trusted, but on the other hand, why not reuse them
for a noble purpose...) -- has anyone done any manipulation of the XML
output of wiktionaries and can offer some pointers on that, please?
Conceptually, it seems rather trivial, but maybe there are some
obstacles that I am not aware of.

Thanks,

  Piotr