Esempio n. 1
0
 def __init__(self, order=3, smoothing=None, tokenize=None):
     if order < 1:
         print("Invalid order: " + str(order)),
         print("Defaulting to order 3.")
         self.order = 3
     self.order = order
     self.counts = dict()
     self.tokenize = tokenize
     if not self.tokenize:
         self.tokenize = DEFAULT_TOKENIZE
     self.Starter = '<S>'
     self.Ender = '</S>'
     self.OOV = '!OOV!'
     self.probEst = smoothing # might be None
     if not self.probEst:
         from probest import MLE
         self.probEst = MLE()