def peter_norvig_approach(self, word):
if word in self.vocab:
return [word]
p_max = 0
approx = []
candidates = known_generated(word, self.vocab)
for c in candidates:
p = self.vocab[c]
if p > p_max:
p_max = p
approx = [c]
elif p == p_max:
approx.append(c)
if approx:
return approx
else:
return [word]
def context_approach_trigrams(self, utterance):
w1, w2, w3 = word_tokenize(utterance)
c_based_on_context = self.tri_grams[w1][w2]
if c_based_on_context:
vocab = {c for c in c_based_on_context}
candidates = known_generated(w3, vocab)
if candidates:
p_max = 0
approx = []
for c in candidates:
p = c_based_on_context[c]
if p_max < p:
p_max = p
approx = [c]
elif p_max == p:
approx.append(c)
return approx
return ["xxxxx"]
def context_approach_bigrams(self, utterance):
w1, w2 = word_tokenize(utterance)
w1 = self.word2idx[w1]
c_based_on_context = self.bi_grams[w1]
if c_based_on_context:
vocab = {self.idx2word[c] for c in c_based_on_context}
candidates = known_generated(w2, vocab)
if candidates:
candidates = tuple([self.word2idx[c] for c in candidates])
p_max = 0
approx = []
for c in candidates:
p = c_based_on_context[c]
if p_max < p:
p_max = p
approx = [c]
elif p_max == p:
approx.append(c)
return tuple([self.idx2word[i] for i in approx])
return ["xxxxx"]
def generate_and_scan(self, word):
candidates = list(known_generated(word, self.vocab))
if candidates:
return candidates
return ["xxxx"]