OK apparently using OpenNLP 1.5.3 works. Something is broken in the
later versions. I'll try to file a bug report.
On 30.07.2018 20:00, Xiang Ji wrote:
This seems to have something to do with the CoNLL2003 format. If I try
to run the trainer directly it works. However my input data is in
CoNLL 2003 format. Running `TokenNameFinderConverter` gives me the
same error. Even trying it on the official example
https://github.com/apache/opennlp/blob/master/opennlp-tools/src/test/resources/opennlp/tools/formats/conll2003-en.sample
doesn't work.
On 30.07.2018 18:33, Xiang Ji wrote:
Hi,
I’m trying to test training OpenNLP’s Name Finder on some data,
according to the guide in the documentation. However, I encountered
the error: |Unsupported language: en|, which doesn’t seem to make any
sense.
The command I ran is: |opennlp TokenNameFinderTrainer.conll03 -model
model.bin -lang en -types per,loc,org,misc -data train.txt -encoding
UTF-8|
I downloaded OpenNLP 1.9.0 from
https://opennlp.apache.org/download.html. The |OPENNLP_HOME|
environment variable does seem to be properly set, and the |lang|
folder in the base folder contains an |en| folder.
Best regards,
Xiang Ji
