This is just a draft up to the point where I would need to ask about the row labels or evaluate how well the nltk parser is doing. In habitus and processors we have things built-in to filter out junk sentences. Something like that may be necessary. It usually requires tokenization of words, which isn't included here yet.
This is just a draft up to the point where I would need to ask about the row labels or evaluate how well the nltk parser is doing. In habitus and processors we have things built-in to filter out junk sentences. Something like that may be necessary. It usually requires tokenization of words, which isn't included here yet.