def test_recognition_1_level(self):
text = u"济南杨铭宇餐饮管理有限公司是由杨先生创办的餐饮企业"
self.gen_word(text)
# vertexs = persion_recognition.recognition(vertexs, word_net_optimum, word_net)
# word_net_optimum = WordNet(text, vertexs=vertexs)
organization_recognition.recognition(self.vertexs, self.word_net_optimum, self.word_net)
vertexs = viterbi(self.word_net_optimum.vertexs)
self.assertIn(Vertex(u"济南杨铭宇餐饮管理有限公司", attribute=u"nt 1"), vertexs)
def test_recognition_1_level(self):
text = u"济南杨铭宇餐饮管理有限公司是由杨先生创办的餐饮企业"
self.gen_word(text)
# vertexs = persion_recognition.recognition(vertexs, word_net_optimum, word_net)
# word_net_optimum = WordNet(text, vertexs=vertexs)
organization_recognition.recognition(self.vertexs,
self.word_net_optimum,
self.word_net)
vertexs = viterbi(self.word_net_optimum.vertexs)
self.assertIn(Vertex(u"济南杨铭宇餐饮管理有限公司", attribute=u"nt 1"), vertexs)
def test_recognition_2_level(self):
person_recognition.recognition(self.vertexs, self.word_net_optimum, self.word_net)
place_recognition.recognition(self.vertexs, self.word_net_optimum, self.word_net)
word_net_optimum = WordNet(self.text, vertexs=self.vertexs)
vertexs = organization_recognition.recognition(self.vertexs, word_net_optimum, self.word_net)
# viterbi(word_net_optimum.vertexs)
dump_vertexs(vertexs)
self.assertIn(Vertex(u"济南杨铭宇餐饮管理有限公司", attribute=u"nt 1"), vertexs)
def test_organization_recognition(self):
text = traditional_to_simplified(u"馬總統上午前往陸軍航空601旅，")
Config.debug = True
self.gen_word(text)
person_recognition.recognition(self.vertexs, self.word_net_optimum, self.word_net)
place_recognition.recognition(self.vertexs, self.word_net_optimum, self.word_net)
word_net_optimum = WordNet(self.text, vertexs=self.vertexs)
vertexs = organization_recognition.recognition(self.vertexs, word_net_optimum, self.word_net)
dump_vertexs(vertexs)
self.assertIn(Vertex(u"陆军航空601旅", attribute=u"nt 1"), vertexs)
def test_organization_recognition(self):
text = traditional_to_simplified(u"馬總統上午前往陸軍航空601旅，")
Config.debug = True
self.gen_word(text)
person_recognition.recognition(self.vertexs, self.word_net_optimum,
self.word_net)
place_recognition.recognition(self.vertexs, self.word_net_optimum,
self.word_net)
word_net_optimum = WordNet(self.text, vertexs=self.vertexs)
vertexs = organization_recognition.recognition(self.vertexs,
word_net_optimum,
self.word_net)
dump_vertexs(vertexs)
self.assertIn(Vertex(u"陆军航空601旅", attribute=u"nt 1"), vertexs)
def test_recognition_2_level(self):
text = u"济南杨铭宇餐饮管理有限公司是由杨先生创办的餐饮企业"
self.gen_word(text)
person_recognition.recognition(self.vertexs, self.word_net_optimum,
self.word_net)
place_recognition.recognition(self.vertexs, self.word_net_optimum,
self.word_net)
word_net_optimum = WordNet(self.text, vertexs=self.vertexs)
vertexs = organization_recognition.recognition(self.vertexs,
word_net_optimum,
self.word_net)
# viterbi(word_net_optimum.vertexs)
dump_vertexs(vertexs)
self.assertIn(Vertex(u"济南杨铭宇餐饮管理有限公司", attribute=u"nt 1"), vertexs)
def seg_to_vertexs(text):
word_net = WordNet(text)
# 粗分词网
gen_word_net(text, word_net)
if Config.debug:
print(u"打印粗分词网：")
print(unicode(word_net))
# 维特比
vertexs = viterbi(word_net.vertexs)
if Config.use_custom_dict:
vertexs = combine_by_custom_dict(vertexs)
word_net_optimum = WordNet(text, vertexs=vertexs)
if Config.name_recognize:
person_recognition.recognition(vertexs, word_net_optimum, word_net)
if Config.place_recognize:
place_recognition.recognition(vertexs, word_net_optimum, word_net)
if Config.debug:
print(u"打印人名、地名识别词网：")
print(unicode(word_net_optimum))
vertexs = viterbi(word_net_optimum.vertexs)
if Config.org_recognize:
word_net_optimum = WordNet(text, vertexs=vertexs)
vertexs = organization_recognition.recognition(vertexs, word_net_optimum, word_net)
if Config.debug:
print(u"打印人组织识别词网：")
print(unicode(word_net_optimum))
return vertexs