-
Possible classification features:
- appearance in a gazetter
- gazetter from nltk
- NER tags
- nltk collocations module
- appearance in a gazetter
-
Add some sort of cache for question classification
-
Extractors:
- handle stop words better, check for contains instead of just equal