def test_ChatClient():
# def __init__(self, language, intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file):
# os.getcwd() - 현재 프로세스의 작업 디렉토리 얻기
# os.curdir() - 현재 디렉토리 얻기
# os.pardir() - 부모 디렉토리 얻기
#특정 경로에 대해 절대 경로 얻기 os.path.abspath(".\\Scripts")
print("현재 프로세스의 작업 디렉토리 [%s]" % os.getcwd())
print("현재 디렉토리[%s]" % os.curdir)
print("부모 디렉토리[%s]" % os.pardir)
print("부모 디렉토리로 변경[%s]" % os.chdir(os.pardir))
# print("변경후 현재 프로세스의 작업 디렉토리 [%s]" % os.getcwd())
print("ArkChatFramework디렉토리 [%s]" % os.path.abspath("ArkChatFramework"))
print("intents 파일 디렉토리[%s]" %
os.path.abspath("ArkNLU/DialogIntents/intents_home_kr.json"))
input_file_name = os.path.abspath(
"ArkNLU/DialogIntents/intents_home_kr.json")
print("intents 파일 디렉토리[%s]" % input_file_name)
input_training_data_file_name = os.path.abspath(
"ArkNLU/NLUModel/training_data_home_kr")
print("training 파일 디렉토리[%s]" % input_training_data_file_name)
tflearn_logs_dir = os.path.abspath('ArkNLU/NLUModel/home_tflearn_kr_logs')
print("tflearn_logs 디렉토리[%s]" % tflearn_logs_dir)
tflearn_model_file_name = os.path.abspath(
'ArkNLU/NLUModel/model_home_kr.tflearn')
print("tflearn_model 파일 디렉토리[%s]" % tflearn_model_file_name)
# intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file
learning_model_files = dict(
intents_file=input_file_name,
training_data_file=input_training_data_file_name,
tflearn_logs_dir=tflearn_logs_dir,
tflearn_model_file=tflearn_model_file_name)
bot = ChatClient('ko-KR', learning_model_files)
print("ChatClient instance...")
# bot.read_dialog_intents_jsonfile()
# bot.restore_training_data_structures()
# bot.restore_training_model()
userID = 'arkwith7'
sentence = '미션이 무엇입니까?'
results = bot.get_classify(sentence)
print("미션이 무엇입니까? 사용자의도분류[%s]" % results)
# 대화 말뭉치와 대화 의도가 정의된 JSON 문서 집합 읽기
print("미션이 무엇입니까? 응답[%s]" %
bot.response(sentence, userID, show_details=True))
def get_answer(self, sentence, userID='123', show_details=False):
results = ChatClient.get_classify(self, sentence)
# if we have a classification then find the matching intent tag
if results:
# loop as long as there are matches to process
while results:
for i in self.dialog_intents['intents']:
# find a tag matching the first result
if i['tag'] == results[0][0]:
# send "no_matching" message when the matching rate of the message is lower than self.criteria_coincidence.
if self.criteria_coincidence > results[0][1]:
return "no_matching"
# set context for this intent if necessary
if 'context_set' in i:
if show_details:
self.logger.debug('context:', i['context_set'])
self.context[userID] = i['context_set']
# check if this intent is contextual and applies to this user's conversation
if not 'context_filter' in i or \
(userID in self.context and 'context_filter' in i and i['context_filter'] == self.context[userID]):
if show_details:
self.logger.debug('tag:', i['tag'])
# a random response from the intent
return random.choice(i['responses'])
results.pop(0)
def get_answer(self, sentence, userID='123', show_details=False):
#If a conversation that is not understood during the chatting is in progress more than 3 times,
# send a usage guide message.
if self.not_matching_count >= 3:
self.not_matching_count = 1
self.previous_not_matching = False
return self.usage_guide_message
results = ChatClient.get_classify(self, sentence)
#Echo processing when using bad language three times in a row.
if self.slang_matching_count >= 3:
if results[0][0] != 'Slang':
self.slang_matching_count = 1
self.previous_slang_matching = False
else:
return sentence
# if we have a classification then find the matching intent tag
if results:
# loop as long as there are matches to process
while results:
for i in self.dialog_intents['intents']:
# find a tag matching the first result
if i['tag'] == results[0][0]:
# send message when the matching rate of the message is lower than self.criteria_coincidence.
if self.criteria_coincidence > results[0][1]:
return_message = '"' + sentence + '"' + self.not_matching_message
if self.previous_not_matching:
self.not_matching_count += 1
self.previous_not_matching = True
return return_message
# check when using bad language three times in a row.
if results[0][0] == 'Slang':
if self.previous_slang_matching:
self.slang_matching_count += 1
self.previous_slang_matching = True
# set context for this intent if necessary
if 'context_set' in i:
if show_details:
self.logger.debug('context:', i['context_set'])
self.context[userID] = i['context_set']
# check if this intent is contextual and applies to this user's conversation
if not 'context_filter' in i or \
(userID in self.context and 'context_filter' in i and i['context_filter'] == self.context[userID]):
if show_details:
self.logger.debug('tag:', i['tag'])
# a random response from the intent
return random.choice(i['responses'])
else:
return self.context_filter_message
results.pop(0)
def __init__(self, work_dir):
'''
Constructor
'''
self.logger = logging.getLogger(__name__)
# logger.info("작업디렉토리 : %s" % os.getcwd())
self.intents_file_name = os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/DialogIntents/intents_catechism_kr.json")
# logger.info("intents 파일 디렉토리[%s]" % input_file_name)
self.input_training_data_file_name = os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/NLUModel/training_data_catechism_kr")
# logger.info("training 파일 디렉토리[%s]" % input_training_data_file_name)
self.tflearn_logs_dir = os.path.join(
work_dir,
'ArkChatFramework/ArkNLU/NLUModel/catechism_tflearn_kr_logs')
# logger.info("tflearn_logs 디렉토리[%s]" % tflearn_logs_dir)
self.tflearn_model_file_name = os.path.join(
work_dir,
'ArkChatFramework/ArkNLU/NLUModel/model_catechism_kr.tflearn')
# logger.info("tflearn_model 파일 디렉토리[%s]" % tflearn_model_file_name)
# intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file
learning_model_files = dict(
intents_file=self.intents_file_name,
training_data_file=self.input_training_data_file_name,
tflearn_logs_dir=self.tflearn_logs_dir,
tflearn_model_file=self.tflearn_model_file_name)
ChatClient.__init__(self, 'ko-KR', learning_model_files)
self.context = {}
self.criteria_coincidence = 0.80
self.not_matching_count = 1
self.previous_not_matching = False
self.slang_matching_count = 1
self.previous_slang_matching = False
self.context_filter_message = "상위 질문 이후 가능합니다."
self.not_matching_message = "에 대해 이해하지 못했습니다."
self.usage_guide_message = "여기서는 음성과 문자 채팅으로 웨스트민스터 소요리 문답의 내용인 하나님에 대하여, 사람에 대하여, 예수그리스도에 대하여, 신앙생활에 대하여와 관련된 내용만 채팅이 가능합니다."
self.previous_context_set = ''
self.previous_responses = ''
def conversation_NLULearning():
print("현재 프로세스의 작업 디렉토리 [%s]" % os.getcwd())
print("부모 디렉토리로 변경[%s]" % os.chdir(os.pardir))
print("변경후 현재 프로세스의 작업 디렉토리 [%s]" % os.getcwd())
print(
"intents 파일 디렉토리[%s]" %
os.path.abspath("ArkNLU/DialogIntents/intents_conversations_kr.json"))
input_file_name = os.path.abspath(
"ArkNLU/DialogIntents/intents_conversations_kr.json")
print("intents 파일 디렉토리[%s]" % input_file_name)
input_training_data_file_name = os.path.abspath(
"ArkNLU/NLUModel/training_data_conversations_kr")
print("training 파일 디렉토리[%s]" % input_training_data_file_name)
tflearn_logs_dir = os.path.abspath(
'ArkNLU/NLUModel/conversations_tflearn_kr_logs')
print("tflearn_logs 디렉토리[%s]" % tflearn_logs_dir)
tflearn_model_file_name = os.path.abspath(
'ArkNLU/NLUModel/model_conversations_kr.tflearn')
print("tflearn_model 파일 디렉토리[%s]" % tflearn_model_file_name)
# intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file
learning_model_files = dict(
intents_file=input_file_name,
training_data_file=input_training_data_file_name,
tflearn_logs_dir=tflearn_logs_dir,
tflearn_model_file=tflearn_model_file_name)
bot = ChatClient('ko-KR', learning_model_files)
print("conversations NLULearning instance...")
if bot.create_learning_model(show_details=True):
print("conversations model creation success....")
else:
print("conversations model creation fail....")
def home_NLULearning(work_dir):
print("현재 프로세스의 작업 디렉토리 [%s]" % os.getcwd())
print("부모 디렉토리로 변경[%s]" % os.chdir(os.pardir))
print("변경후 현재 프로세스의 작업 디렉토리 [%s]" % os.getcwd())
print("intents 파일 디렉토리[%s]" % os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/DialogIntents/intents_home_kr.json"))
input_file_name = os.path.join(
work_dir, "ArkChatFramework/ArkNLU/DialogIntents/intents_home_kr.json")
print("intents 파일 디렉토리[%s]" % input_file_name)
input_training_data_file_name = os.path.join(
work_dir, "ArkChatFramework/ArkNLU/NLUModel/training_data_home_kr")
print("training 파일 디렉토리[%s]" % input_training_data_file_name)
tflearn_logs_dir = os.path.join(
work_dir, 'ArkChatFramework/ArkNLU/NLUModel/home_tflearn_kr_logs')
print("tflearn_logs 디렉토리[%s]" % tflearn_logs_dir)
tflearn_model_file_name = os.path.join(
work_dir, 'ArkChatFramework/ArkNLU/NLUModel/model_home_kr.tflearn')
print("tflearn_model 파일 디렉토리[%s]" % tflearn_model_file_name)
# intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file
learning_model_files = dict(
intents_file=input_file_name,
training_data_file=input_training_data_file_name,
tflearn_logs_dir=tflearn_logs_dir,
tflearn_model_file=tflearn_model_file_name)
bot = ChatClient('ko-KR', learning_model_files)
print("home NLULearning instance...")
if bot.create_learning_model(show_details=True):
print("home model creation success....")
else:
print("home model creation fail....")
def __init__(self, work_dir):
'''
Constructor
'''
self.logger = logging.getLogger(__name__)
# logger.info("작업디렉토리 : %s" % os.getcwd())
self.intents_file_name = os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/DialogIntents/intents_home_ishift_kr.json"
)
# logger.info("intents 파일 디렉토리[%s]" % input_file_name)
self.input_training_data_file_name = os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/NLUModel/ishift/training_data_home_kr")
# logger.info("training 파일 디렉토리[%s]" % input_training_data_file_name)
self.tflearn_logs_dir = os.path.join(
work_dir,
'ArkChatFramework/ArkNLU/NLUModel/ishift/home_tflearn_kr_logs')
# logger.info("tflearn_logs 디렉토리[%s]" % tflearn_logs_dir)
self.tflearn_model_file_name = os.path.join(
work_dir,
'ArkChatFramework/ArkNLU/NLUModel/ishift/model_home_kr.tflearn')
# logger.info("tflearn_model 파일 디렉토리[%s]" % tflearn_model_file_name)
# intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file
learning_model_files = dict(
intents_file=self.intents_file_name,
training_data_file=self.input_training_data_file_name,
tflearn_logs_dir=self.tflearn_logs_dir,
tflearn_model_file=self.tflearn_model_file_name)
ChatClient.__init__(self, 'ko-KR', learning_model_files)
self.context = {}
self.criteria_coincidence = 0.50
self.not_matching_count = 1
self.previous_not_matching = False
self.slang_matching_count = 1
self.previous_slang_matching = False
self.context_filter_message = "솔루션 에 대한 문의 이후에 관련 기능에 대한 설명이 가능합니다."
self.not_matching_message = "에 대해 이해하지 못했습니다."
self.usage_guide_message = "여기서는 음성과 문자 채팅으로 아이시프트의 vision, mission, 제품, 서비스와 관련된 내용만 채팅이 가능합니다."
def __init__(self, work_dir):
'''
Constructor
'''
self.logger = logging.getLogger(__name__)
# logger.info("작업디렉토리 : %s" % os.getcwd())
self.intents_file_name = os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/DialogIntents/intents_conversations_kr.json"
)
# logger.info("intents 파일 디렉토리[%s]" % input_file_name)
self.input_training_data_file_name = os.path.join(
work_dir,
"ArkChatFramework/ArkNLU/NLUModel/training_data_conversations_kr")
# logger.info("training 파일 디렉토리[%s]" % input_training_data_file_name)
self.tflearn_logs_dir = os.path.join(
work_dir,
'ArkChatFramework/ArkNLU/NLUModel/conversations_tflearn_kr_logs')
# logger.info("tflearn_logs 디렉토리[%s]" % tflearn_logs_dir)
self.tflearn_model_file_name = os.path.join(
work_dir,
'ArkChatFramework/ArkNLU/NLUModel/model_conversations_kr.tflearn')
# logger.info("tflearn_model 파일 디렉토리[%s]" % tflearn_model_file_name)
# intents_file, training_data_file, tflearn_logs_dir, tflearn_model_file
learning_model_files = dict(
intents_file=self.intents_file_name,
training_data_file=self.input_training_data_file_name,
tflearn_logs_dir=self.tflearn_logs_dir,
tflearn_model_file=self.tflearn_model_file_name)
ChatClient.__init__(self, 'ko-KR', learning_model_files)
self.context = {}
self.criteria_coincidence = 0.50
self.not_matching_count = 1
self.previous_not_matching = False
self.slang_matching_count = 1
self.previous_slang_matching = False
def get_answer(self, sentence, userID='123', show_details=False):
#If a conversation that is not understood during the chatting is in progress more than 3 times,
# send a usage guide message.
if self.not_matching_count >= 3:
self.not_matching_count = 1
self.previous_not_matching = False
return self.set_answer("abnormal", "guide_message",
self.usage_guide_message, "")
results = ChatClient.get_classify(self, sentence)
#Echo processing when using bad language three times in a row.
if self.slang_matching_count >= 3:
if results[0][0] != 'Slang':
self.slang_matching_count = 1
self.previous_slang_matching = False
else:
return self.set_answer("abnormal", "not_matching", sentence,
"")
# if we have a classification then find the matching intent tag
if results:
# loop as long as there are matches to process
while results:
for i in self.dialog_intents['intents']:
# find a tag matching the first result
if i['tag'] == results[0][0]:
self.logger.debug("previous_context_set : [%s] ",
self.previous_context_set)
# send message when the matching rate of the message is lower than self.criteria_coincidence.
if self.criteria_coincidence > results[0][1]:
return_message = '"' + sentence + '"' + self.not_matching_message
if self.previous_not_matching:
self.not_matching_count += 1
self.previous_not_matching = True
return self.set_answer("abnormal", "not_matching",
return_message, "")
# check when using bad language three times in a row.
if results[0][0] == 'Slang':
if self.previous_slang_matching:
self.slang_matching_count += 1
self.previous_slang_matching = True
# set context for this intent if necessary
if 'context_set' in i:
if show_details:
self.logger.debug('context:', i['context_set'])
self.context[userID] = i['context_set']
# check context_set for this intent
if i['context_set']:
self.previous_context_set = i['context_set']
self.previous_responses = i['responses']
self.logger.debug("context_set : [%s] ",
i['context_set'])
response = {}
response["response"] = i['responses']
return self.set_answer("list", i['patterns'][0],
i['responses'], "")
# check if this intent is contextual and applies to this user's conversation
if not 'context_filter' in i or \
(userID in self.context and 'context_filter' in i and i['context_filter'] == self.context[userID]):
if show_details:
self.logger.debug('tag:', i['tag'])
# a random response from the intent
self.previous_context_set = i['context_set']
self.previous_responses = i['responses']
explanation = []
if ('responses_ref' in i):
explanation = i['responses_ref']
return self.set_answer(
"qna", i['patterns'][0],
random.choice(i['responses']), explanation)
elif self.previous_context_set == i['context_filter']:
self.logger.debug("context_set : [%s] ",
i['context_set'])
self.logger.debug("context_filter : [%s] ",
i['context_filter'])
self.logger.debug("previous_context_set : [%s] ",
self.previous_context_set)
self.previous_context_set = i['context_set']
self.previous_responses = i['responses']
return self.set_answer("list", i['patterns'][0],
i['responses'], "")
else:
if not self.previous_responses:
self.previous_responses = i['context_filter']
return_values = i['context_filter']
else:
return_values = ""
for value in self.previous_responses:
return_values += value + ' '
self.previous_responses = i['context_filter']
# context_filter와 같은 tag의 responses값을 보여 주고 선택하게 함
return self.set_answer(
"abnormal", "not_matching", return_values +
"와 같은 " + self.context_filter_message, "")
results.pop(0)