On 29 October 2010 11:25, Antonio Toral <[email protected]> wrote:
> Hi Jim, Fran,
>
>> >> - en.metadix:
>> >> <pardef n="Aachen__np"><e><p><l/><r><s n="np"/><s
>> >> n="sg"/></r></p></e></pardef>
>> >> <e lm="Yekaterinburg"><i>Yekaterinburg</i><par n="Aachen__np"/></e>
>> >>
>> >
>> > See, this is np.loc, because it's a place. If you want a generic, 'I
>> > don't know what this is' np, use np.al
>
> thanks for your feedback. I've tried with np.al and using existing
> pardefs intead of my "funky homegrown" ones... and it seems to work...
> i'm still getting <SN><UNDET> in en->es transfer but the final output
> looks clean ;)
>
No, that's correct -- SN because it's a noun phrase, UNDET because
it's not used with a definite determiner.
>
> en.metadix:
> <pardef n="Linux__np"><e><p><l/><r><s n="np"/><s n="al"/><s
> n="sg"/></r></p></e></pardef>
> <e lm="Yekaterinburg"><i>Yekaterinburg</i><par n="Linux__np"/></e>
>
> es.dix:
> <pardef n="ABC__np"><e><p><l/><r><s n="np"/><s n="al"/><s n="m"/><s
> n="sg"/></r></p></e></pardef>
> <e lm="Ekaterimburgo"><i>Ekaterimburgo</i><par n="ABC__np"/></e>
>
> en-es.dix:
> <e><p><l>Yekaterinburg<s n="np"/><s n="al"/></l><r>Ekaterimburgo<s
> n="np"/><s n="al"/><s n="m"></r></p></e>
>
> # echo "Yekaterinburg" | apertium-destxt | lt-proc en-es.automorf.bin |
> apertium-tagger -g en-es.prob | apertium-pretransfer | apertium-transfer
> apertium-en-es.en-es.t1x en-es.t1x.bin en-es.autobil.bin
> ^nom<SN><UNDET><m><sg>{^Ekaterimburgo<np><al><3><4>$}$^punt<sent>{^.<sent>$}$[][
>
> # echo "Ekaterimburgo" | apertium-destxt | lt-proc es-en.automorf.bin |
> apertium-tagger -g es-en.prob | apertium-pretransfer | apertium-transfer
> apertium-en-es.es-en.t1x es-en.t1x.bin es-en.autobil.bin
> ^Al<SN><m><sg>{^Yekaterinburg<np><al><sg>$}$^punt<sent>{^.<sent>$}$[][
>
> # echo "Yekaterinburg" | apertium -d . test-en-es
> Ekaterimburgo
>
>
> btw, "make clean" does not delete the files generated under .deps/
> (en-es-0.7.1)
>
I think I fixed that in SVN not too long ago.
>> Actually, if you can contribute the wordlist, send it to me and I'll
>> subcategorise it. Outsource the problem :)
>
> yep, my idea is to contribute the output (if valuable at all), but
> before I have to "stabilise" the procedures ;)
>
Well, I've been meaning to add a script to check DBpedia for those
sorts of things, but at the moment I have a horrible set of scripts
for checking against a wikipedia dump that I wouldn't want anyone to
see :)
--
<Leftmost> jimregan, that's because deep inside you, you are evil.
<Leftmost> Also not-so-deep inside you.
------------------------------------------------------------------------------
Nokia and AT&T present the 2010 Calling All Innovators-North America contest
Create new apps & games for the Nokia N8 for consumers in U.S. and Canada
$10 million total in prizes - $4M cash, 500 devices, nearly $6M in marketing
Develop with Nokia Qt SDK, Web Runtime, or Java and Publish to Ovi Store
http://p.sf.net/sfu/nokia-dev2dev
_______________________________________________
Apertium-stuff mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/apertium-stuff
