def pre_process(self, text):
if self.no_dictionary:
x_processed = tu.normalize_punctuation(text).split()
else:
x_processed = [word for word in tu.normalize_punctuation(text).split()
if word in self.dictionary.token2id and word not in self.stoplist]
return x_processed
def fit(self, X, y=None):
logging.info("Loaded from file")
x_clean = [tu.normalize_punctuation(text).split() for text in X]
if self.w2v_model is None:
self.w2v_model = w2v_models.build_word2vec(x_clean, size=100, window=10, min_count=1, dataname="test")
if self.no_below == 1 and self.no_above == 1:
self.no_dictionary = True
else:
self.dictionary = corpora.Dictionary(x_clean)
self.dictionary.filter_extremes(no_above=self.no_above, no_below=self.no_below)
# setting the coordinates for different models (start, stop)
size = self.w2v_model.layer1_size
self.feature_crd = {'00_avg': (0, size),
'01_std': (size, 2*size)}
feature_cnt = 2
start = 2*size
l = size
for i in range(1,self.diffmax0):
#name = "%02d_diff0_%i" % (feature_cnt, i)
#feature_cnt += 1
#val = (start, start + l)
#self.feature_crd[name] = val
#start += l
name = "%02d_diff0_std_%i" % (feature_cnt, i)
feature_cnt += 1
val = (start, start + l)
self.feature_crd[name] = val
start += l
for i in range(1,self.diffmax1):
name = "%02d_diff1_%i" % (feature_cnt, i)
feature_cnt += 1
val = (start, start + l)
self.feature_crd[name] = val
start += l
name = "%02d_diff1_std_%i" % (feature_cnt, i)
feature_cnt += 1
val = (start, start + l)
self.feature_crd[name] = val
start += l
self.length = start
logging.info("Total feature length %i " % self.length )
logging.info("W2V: got a model %s " % (self.w2v_model,))
return self
def transform(self, X):
# Text pre-processing
x_clean = [tu.normalize_punctuation(text).split() for text in X]
logging.info("DPGGM: Text prepocessed")
# Vectorize using W2V model
if self.dpgmm is not None:
logging.info("Vectorizing a corpus")
size = self.w2v_model.layer1_size
if len(X) > 0:
vecs = np.concatenate([self.clusterize(z) for z in x_clean], axis=0)
else:
vecs = np.zeros(size).reshape((1, size))
logging.info("DPGMM: returning pre-processed data of shape %s" % (vecs.shape, ))
else:
logging.info("W2V Averaged: no model was provided.")
vecs = np.zeros((len(X), 1))
return vecs
