def __init__(self):
self.MODULE_PATH = pales_index()
self.HEADERS = Config.values()['headers']
dataset_path = os.path.normpath(pales_index() + os.path.normcase("/data/dataset.csv"))
dataset = pandas.read_csv(dataset_path, names=self.HEADERS)
array = dataset.values
self.X = array[:, 0:(len(self.HEADERS) - 1)]
self.Y = array[:, (len(self.HEADERS) - 1)]
self.validation_size = 0.20
def translate_to_keras(url: str) -> numpy.ndarray:
info = dict()
for head in Config.values()['headers']:
info[head] = 0
info.pop('result', None)
data = score(url)
for key in data['comparators'].keys():
info[key] = data['comparators'][key]
for key in data['semantic'].keys():
info[key] = data['semantic'][key]
return numpy.array([[v for k, v in info.items()]])
def getCache(url: str) -> Union[dict, bool]:
hashed = hashlib.sha224(str(url).encode('utf-8')).hexdigest()
cache = os.path.normpath(jano_index() + os.path.normcase("/cache/{0}.jcache".format(hashed)))
if os.path.isfile(cache):
time = pendulum.from_timestamp(os.path.getmtime(cache))
now = pendulum.now(Config.values()['timezone'])
if now.diff(time).in_days() >= 2:
os.remove(cache)
return False
return pickle.load(open(cache, "rb"))
else:
return False
def polaridade(texto: str) -> dict:
"""
Verifica a polaridade de um texto (sentimentos bons/ruins)
:param texto: Texto a ser analisado
:return: Polaridade em número
"""
resultado = dict(good=0, bad=0)
if len(texto) == 0:
return dict()
polyglot_text = Text(filter_stopwords(texto), hint_language_code=Config().values()['language'][:2])
for w in polyglot_text.words:
pol = w.polarity
if pol < 0:
resultado['bad'] += 1
elif pol > 0:
resultado['good'] += 1
resultado['bad'] = round(resultado['bad'] / len(polyglot_text.words), 3)
resultado['good'] = round(resultado['good'] / len(polyglot_text.words), 3)
return resultado
def gramatica(texto: str) -> dict:
"""
Analisa e conta cada token de um texto no formato explicado aqui: http://polyglot.readthedocs.io/en/latest/POS.html
:param texto: Texto a ser analisado
:return: Dicionário com as tags presentes e a quantia delas
"""
resposta = dict()
if len(texto) == 0:
return dict()
polyglot_text = Text(texto, hint_language_code=Config().values()['language'][:2])
for word, tag in polyglot_text.pos_tags:
if tag in resposta.keys():
resposta[tag] += 1
else:
resposta[tag] = 1
# Porcentagem
total = sum(resposta.values())
for tag in resposta.keys():
resposta[tag] = round(resposta[tag] / total, 3)
return resposta
def __init__(self):
self.MODULE_PATH = pales_index()
self.HEADERS = Config.values()['headers']
dataset_path = os.path.normpath(pales_index() + os.path.normcase("/data/dataset.csv"))
dataset = pandas.read_csv(dataset_path, names=self.HEADERS)
array = dataset.values
self.X = array[:, 0:(len(self.HEADERS) - 1)]
self.Y = array[:, (len(self.HEADERS) - 1)]
self.p = {'lr': (0.5, 5, 8),
'first_neuron': [8, 16, 32, 64],
'hidden_layers': [2, 3, 4, 5],
'batch_size': (1, 5, 5),
'epochs': [50, 100, 150],
'dropout': (0, 0.2, 0.5),
'weight_regulizer': [None],
'emb_output_dims': [None],
'shape': ['brick', 'long_funnel'],
'kernel_initializer': ['uniform', 'normal'],
'optimizer': [Adam, Nadam],
'losses': [binary_crossentropy],
'activation': [relu, elu],
'last_activation': [sigmoid]}
def count_words(texto: str) -> int:
polyglot_text = Text(filter_stopwords(texto),
hint_language_code=Config.values()['language'][:2])
return len(polyglot_text.words)