def __init__(self, mode='train'):
ClassicalFeature.__init__(self, mode)
self.extract_stage = [self._extract]
# uni_words = ['what', 'whi', 'which', 'how', 'where', 'when', 'if', 'can', 'should']
# do_words = ['doe', 'do', 'did']
# be_words = ['is', 'are']
# will_words = ['will', 'would']
# self.words = uni_words + do_words + be_words + will_words
self.words = ['what', 'whi', 'which', 'how', 'where', 'when']
self.columns = [
'_'.join(word)
for word in combinations_with_replacement(self.words, 2)
]
def __init__(self, gtype, mode='train'):
ClassicalFeature.__init__(self, mode)
if gtype == 'concurrence':
self.seq2id, self.graph = UnDirectGraph.load()
else:
self.seq2id, self.graph = UnDirectWeightGraph.load()
def __init__(self, mode='train'):
ClassicalFeature.__init__(self, mode)
self.idf = TFIDF.load()
self.extract_stage = [self._extract]
self.idf_vocab = self.idf.get_feature_names()
def __init__(self, mode='train'):
ClassicalFeature.__init__(self, mode)
self.idf = TFIDF.load()
self.extract_stage = [self._extract]
def _init_predict_corpus(self):
self.x_submission = ClassicalFeature.load(mode='test')
def _init_train_corpus(self):
self.x = ClassicalFeature.load()
self.y = pd.read_csv(config.origin_train_file)['is_duplicate']
def __init__(self, mode='train'):
ClassicalFeature.__init__(self, mode)
self.extract_stage = [self._extract]
self.model = W2V.load()
def __init__(self, mode='train'):
ClassicalFeature.__init__(self, mode)
self.extract_stage = [self._init_prob, self._extract]
self.filter_words = list(Punctuations.PUNCTUATIONS) + list(
StopWords.StopWordsEN)
def __init__(self, mode='train'):
ClassicalFeature.__init__(self, mode)
self.extract_stage = [self.init_idf, self._extract]
