On 9/5/11 11:17 AM, Gerber Daniel wrote:
I was just wondering if there is an up to date trained Chunker for the German
language available. It seems, that on the model website [1] there are only
available for the English language. If not, is there a way to train it myself?
I also found this blog post [2] which provided a Chunker for german, but this
is unfortunately not compatible with the current OpenNLP version and the
corpus used to train the data not available to public audience:(
Any ideas?
They trained on the tiger corpus, which you can access for free as far
as I know. Would
be nice to have a little sandbox project where we share training code
for it.
I suggest that you download the english chunker model from our website,
unzip it,
replace the model in it with the model from their website (you must
ungzip it before),
then you should change the language in the contained manifest from "en"
to "de" and
zip everything again.
Jörn
