A few things already changed or easier now compared to when
Jason did write the newlang tutorial.
I suggest that we change it into instructions on how to train OpenNLP
on Conll06 data.
Any opinions ?
If it has become simpler, it is better to make a new one.
I wish for a more abstract explanation allowing people to add any new language from any data for instance with requirements and format used by OpenNLP.
A more general description about the training should be part of the documentation for the component. The newlang tutorial should in my opinion only describe how to convert the conll06 data into the native OpenNLP format.
Actually Jason did write this tutorial, and then it was really easy to see that all the things
he explained could be integrated into OpenNLP and I started working on that. Now
its released and I think we should also rewrite the tutorial.
Would that be ok for you Jason, or do you want to keep it in
the current form ?
By all means, please do rewrite the tutorial to be more appropriate/updated! I really just wanted to get the ball rolling, and I'm very happy with where it has gone and is going.
Log in to post a comment.