Thanks! We've actually built our own probabilistic information extractor, so we don't use the POS tagging, so NLTK is sorta redundant for the approach we took (

Thanks! We've actually built our own probabilistic information extractor, so we don't use the POS tagging, so NLTK is sorta redundant for the approach we took (and also less accurate)