def test_cond_prob_2gram(self):
ngram = NGram(2, self.sents)
probs = {
('pescado', 'come'): 0.5,
('salmón', 'come'): 0.5,
('salame', 'come'): 0.0,
}
for (token, prev), p in probs.items():
self.assertEqual(ngram.cond_prob(token, [prev]), p)
def test_cond_prob_1gram(self):
ngram = NGram(1, self.sents)
probs = {
'pescado': 1 / 12.0,
'come': 2 / 12.0,
'salame': 0.0,
}
for token, p in probs.items():
self.assertEqual(ngram.cond_prob(token), p)
def test_cond_prob_2gram(self):
ngram = NGram(2, self.sents)
probs = {
('pescado', 'come'): 0.5,
('salmón', 'come'): 0.5,
('salame', 'come'): 0.0,
}
for (token, prev), p in probs.items():
self.assertAlmostEqual(ngram.cond_prob(token, (prev,)), p)
def test_cond_prob_1gram(self):
ngram = NGram(1, self.sents)
probs = {
'pescado': 1 / 12.0,
'come': 2 / 12.0,
'salame': 0.0,
}
for token, p in probs.items():
self.assertAlmostEqual(ngram.cond_prob(token), p)
def test_cond_prob_3gram(self):
ngram = NGram(3, self.sents)
probs = {
('pescado', ('gato', 'come')): 1,
('salmón', ('gata', 'come')): 1,
('salame', ('gato', 'come')): 0.0,
('gato', ('<s>', 'el')): 1,
('gata', ('<s>', 'el')): 0.0,
}
for (token, prev), p in probs.items():
self.assertEqual(ngram.cond_prob(token, list(prev)), p)
def test_cond_prob_3gram(self):
ngram = NGram(3, self.sents)
probs = {
('pescado', 'gato', 'come'): 1,
('salmón', 'gata', 'come'): 1,
('salame', 'gato', 'come'): 0.0,
('salmón', 'gato', 'come'): 0.0,
('el', '<s>', '<s>'): 0.5,
('</s>', 'pescado', '.'): 1,
}
for (token, prev1, prev2), p in probs.items():
self.assertEqual(ngram.cond_prob(token, [prev1, prev2]), p)
def test_cond_prob_3gram(self):
ngram = NGram(3, self.sents)
probs = {
('pescado', 'gato', 'come'): 1,
('salmón', 'gata', 'come'): 1,
('salame', 'gato', 'come'): 0.0,
('salmón', 'gato', 'come'): 0.0,
('el', '<s>', '<s>'): 0.5,
('</s>', 'pescado', '.'): 1,
}
for (token, prev1, prev2), p in probs.items():
self.assertEqual(ngram.cond_prob(token, [prev1, prev2]), p)
def test_cond_prob_4gram(self):
ngram = NGram(4, self.sents2)
probs = {
('pescado', ('el', 'gato', 'come')): 1,
('salmón', ('la', 'gata', 'come')): 0.0,
('salame', ('el', 'gato', 'come')): 0.0,
('gato', ('<s>', '<s>', 'el')): 1,
('viejo', ('gato', 'come', 'pescado')): 0.5,
('fresco', ('gato', 'come', 'pescado')): 0.5,
}
for (token, prev), p in probs.items():
self.assertEqual(ngram.cond_prob(token, list(prev)), p)
