def init_tagger(self):
try:
print 'loading tagger...'
self.tagger = Serializer.get_object(self.ser_path)
print 'DONE...'
if self.tagger == False:
print 'tagger not found in disk. Training one'
self.tagger = self.train_classifier_tagger()
print 'saving tagger to disk'
Serializer.save_object(self.tagger, self.ser_path)
print 'DONE...'
except Exception:
print Exception
def load_data(self,new=0):
if(new == 0):
try:
self.list_reviews = Serializer.get_object('../data/reviews.dat')
except:
print 'impossible to load'
elif new == 2:
self.list_reviews = self.create_reviews([0 for n in range(NUM_RANDOM)], [1 for n in range(NUM_RANDOM)])
else:
self.parser = Parser(self.lang)
self.grammar = FeaturesGrammar()
self.tagger = NGramTagger(self.lang)
url = "http://www.booking.com/reviewlist.en.html"
self.data_pos, self.data_neg = self.parser.get_from_url(url,50)
self.list_reviews = self.create_reviews(self.data_pos, self.data_neg)
self.save_reviews()
def load_data(self, new=0):
"""
:method:: load_data
:param new: new == 0 try to load list already created, else create a new one.
"""
if new == 0:
try:
self.list_reviews = Serializer.get_object("../data/reviews.dat")
except:
print "impossible to load"
elif new == 2:
self.list_reviews = self.create_reviews([0 for n in range(NUM_RANDOM)], [1 for n in range(NUM_RANDOM)])
else:
self.parser = Parser(self.lang)
self.grammar = FeaturesGrammar()
self.tagger = NGramTagger(self.lang)
url = "http://www.booking.com/reviewlist.en.html"
self.data_pos, self.data_neg = self.parser.get_from_url(url, 50)
self.list_reviews = self.create_reviews(self.data_pos, self.data_neg)
self.save_reviews()
def save_reviews(self):
try:
Serializer.save_object(self.list_reviews, '../data/reviews.dat')
except:
print 'i/o error. Not saving'
