def unknowWordsSetZero(self, file_name='test.txt'):
'''
统计未登录词，放入word_dict中，value = 0
:param file_name: 测试文件路径
:return:
'''
test_word_dict = calcuBigramWordDistri(file_name)
unknow_word_cnt = 0
for front_word in test_word_dict:
for word in test_word_dict[front_word]:
if not self.inGram(front_word, word):
self.Bigram[front_word][word] = 0.
unknow_word_cnt += 1
print('unknow words count number: {0}, set unknow word value = {1}'.format(
unknow_word_cnt, 0.0))
def __init__(self):
# 继承Dictionary子类
DictionarySmooth.__init__(self)
self.Bigram = calcuBigramWordDistri()
self.unknowWordsSetZero()
self.alpha = self.wittenBellSmoothing()
def __init__(self):
# 继承Dictionary子类
Dictionary.__init__(self)
self.Bigram = calcuBigramWordDistri()