def getRealCosts(corpus):
""" Retorna as funcoes de custo unigrama, bigrama e possiveis fills obtidas a partir do corpus."""
_realUnigramCost, _realBigramCost, _possibleFills = None, None, None
if _realUnigramCost is None:
print('Training language cost functions [corpus: ' + corpus + ']... ')
_realUnigramCost, _realBigramCost = util.makeLanguageModels(corpus)
if corpus == "corpus.txt":
_possibleFills = util.makeInverseRemovalDictionary(corpus, 'aeiou')
else:
_possibleFills = util.makeInverseRemovalDictionary(
corpus, 'aáàãâeéêiíoõôóuú')
print('Done!')
return _realUnigramCost, _realBigramCost, _possibleFills
def getRealCostspt(corpus='corpuspt.txt'):
""" Retorna as funcoes de custo unigrama, bigrama e possiveis fills obtidas a partir do corpus em portugues."""
_realUnigramCost, _realBigramCost, _possibleFills = None, None, None
if _realUnigramCost is None:
print('Treinamento em portugues do custo da funcão [corpus: ' +
corpus + ']... ')
_realUnigramCost, _realBigramCost = util.makeLanguageModels(corpus)
_possibleFills = util.makeInverseRemovalDictionary(corpus, 'aeiou')
print('Done!')
return _realUnigramCost, _realBigramCost, _possibleFills