Hello,
I'll gladly help you with this project. I haven't used NLTK trainer specifically,
but have experience with NLTK in one of my past projects. The job was to
recognize Russian contact names and postal addresses from people's resumes.
Postal addresses can be written in enormous count of variations, so that job required
some machine learning approach.
Your SO thread describes the issue in great detail. Coud you please
give me actual piece of code? I'll try to replicate your environment and
see if I get same error.
Thank you!
Best regards,
Max Dargan