def __init__(self, min_len=2, min_occ=30): Transformer.__init__(self) self.min_len = min_len self.min_occ = min_occ self.no_trash_wrds = None self.map = lambda doc: [ filter(lambda wrd: wrd in self.no_trash_wrds, sent) for sent in doc ]
def __init__(self, window_width=5, sigma=0.5, min_occur=1, min_word_len=3, top=1000, do_apply=True): Transformer.__init__(self) self.window_width = window_width self.sigma = sigma self.min_occur = min_occur self.min_word_len = min_word_len self.top = top self.do_apply = do_apply