Improving how clustering label is picked

[websorz]

consider stemming the label so it is more representative of the singular entity.

an example:
from nltk.stem import PorterStemmer
ps = PorterStemmer()
ps.stem('Donald Trumps')

out: 'donald trump'