def main():
with open(prior_property_file) as json_file:
prior_property_dict = json.load(json_file)
dialogDB_json = {}
input_json = request.get_json()
print(input_json)
user_ID = input_json['user_id']
userDB_json = {'userID': user_ID, 'command': 'LOGIN'}
userDB_response = Dialog_Manager.UserDBaccess(userDB_json)
KB_language = {
'user_ID': user_ID,
'entities': [],
'frames': [],
'triples': []
}
result_json = {
'user_id': user_ID,
'frames': [],
'entities': [],
'q_list': [],
'knowledge': []
}
user_input = input_json['utterance']
dialogDB_json['user_id'] = KB_language['user_ID']
dialogDB_json['utterance'] = user_input
dialogDB_json['speaker'] = 'user'
dialogDB_json['mode'] = 'make_table'
dialog_index = DialogDBaccess(dialogDB_json)
dialogDB_json['mode'] = 'add_data'
dialogDB_json['utterance'] = user_input
dialogDB_json['speaker'] = 'user'
dialog_index = DialogDBaccess(dialogDB_json)
if not input_json['q_list'] and not input_json['knowledge']:
NLU_response = NLU.Frame_Interpreter(user_input)
#frame기반
if NLU_response['frames']:
KB_language['frames'] = NLU_response['frames']
result_json['frames'] = KB_language['frames']
now_frame = KB_language['frames'][-1]
question_list = []
with open('./frame_info_full.json', 'r', encoding='utf-8') as f:
frame_json_data = json.load(f)
frame_data = frame_json_data[now_frame['frame']]
for ele in frame_data['arguments']:
if ele['coreType'] == 'Core':
question_list.append(ele['fe'])
for exi_ele in now_frame['ele']:
if exi_ele in question_list:
question_list.remove(exi_ele)
if question_list:
result_json['q_list'] = question_list
system_output = \
NLG([now_frame['frame'] + '\t' + question_list[0] + '\t' + '?o'],
'Knowledge_question')[0]
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
system_output = '감사합니다.'
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
#entity기반
elif NLU_response['entities']:
KB_language['entities'] = NLU_response['entities']
result_json['entities'] = KB_language['entities']
KBM_response = Entity_Summarization.ES(
KB_language['entities'][0]['text'])
KB_language['triples'] = KBM_response
utterances = NLG(KB_language['triples'], 'Knowledge_inform')
system_output = ''
for candi in utterances:
if 'wiki' in candi or 'abstract' in candi:
continue
system_output += candi
tmp_list = []
for ele in NLU_response['entities'][0]['type']:
if 'http://dbpedia.org/ontology/' in ele:
tmp_list.append(ele.split('/')[-1])
KB_language['entities'][0]['type'] = copy.deepcopy(tmp_list)
for tmp_class in KB_language['entities'][0]['type']:
if tmp_class in class_dict['level_4']:
entity_class = tmp_class
break
elif tmp_class in class_dict['level_3']:
entity_class = tmp_class
break
elif tmp_class in class_dict['level_2']:
entity_class = tmp_class
break
elif tmp_class in class_dict['level_1']:
entity_class = tmp_class
break
question_list = []
question_property_list = prior_property_dict[entity_class]
question_num = 0
for candi_question in question_property_list:
if question_num == 3:
break
tmp_user_query = Dialog_Manager.SPARQL_Generation(
'ASK',
[KB_language['entities'][0]['uri'], candi_question, '?o'],
KB_language['user_ID'])
tmp_master_query = Dialog_Manager.SPARQL_Generation(
'ASK',
[KB_language['entities'][0]['uri'], candi_question, '?o'],
"")
userDB_json['query'] = tmp_user_query
userDB_json['command'] = 'QUERY'
#print(UserDBaccess(userDB_json))
if not Dialog_Manager.MasterDBaccess(
tmp_master_query) and not Dialog_Manager.UserDBaccess(
userDB_json)['query_result']:
question_list.append([
KB_language['entities'][0]['text'], candi_question,
'?o'
])
question_num += 1
if question_list:
result_json['q_list'] = question_list
system_output += '\n' + KB_language['entities'][0][
'text'] + '에 대해서 몇 가지 물어보고 싶은게 있어요.'
system_output += '\n' + NLG([
question_list[0][0] + '\t' + question_list[0][1] + '\t' +
question_list[0][2]
], 'Knowledge_question')[0]
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
system_output = '감사합니다.'
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
system_output = '제가 아직 이해할 수 없는 문장이에요.'
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
elif input_json['q_list']:
dialogDB_json['utterance'] = user_input
dialogDB_json['speaker'] = 'user'
dialog_index = DialogDBaccess(dialogDB_json)
if input_json['frames']:
result_json['frames'] = input_json['frames']
tmp_user_answer = ETRI_NER(user_input)
tmp_knowledge_list = input_json['knowledge']
if tmp_user_answer:
tmp_knowledge_list.append(tmp_user_answer[0][0])
result_json['knowledge'] = tmp_knowledge_list
result_json['q_list'] = input_json['q_list']
del result_json['q_list'][0]
if result_json['q_list']:
now_frame = input_json['frames'][-1]
system_output = \
NLG([now_frame['frame'] + '\t' + result_json['q_list'][0] + '\t' + '?o'],
'Knowledge_question')[0]
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
if input_json['knowledge']:
#frame지식저장
system_output = '감사합니다'
result_json['frames'] = []
result_json['entities'] = []
result_json['knowledge'] = []
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
system_output = '감사합니다'
result_json['frames'] = []
result_json['entities'] = []
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
elif input_json['entities']:
result_json['entities'] = input_json['entities']
tmp_user_answer = ETRI_NER(user_input)
new_triple = []
tmp_knowledge_list = input_json['knowledge']
result_json['q_list'] = input_json['q_list']
if tmp_user_answer:
new_triple.append('http://kbox.kaist.ac.kr/resource/' +
result_json['q_list'][0][0] + '\t' +
result_json['q_list'][0][1] +
'\thttp://kbox.kaist.ac.kr/resource/' +
tmp_user_answer[0][0])
new_triple.append(
'http://kbox.kaist.ac.kr/resource/' +
result_json['q_list'][0][0] + result_json['q_list'][0][1] +
'http://kbox.kaist.ac.kr/resource/' +
tmp_user_answer[0][0] +
'\thttp://kbox.kaist.ac.kr/flagship/dialogid\thttp://ko.dbpedia.org/resource/'
+ str(dialog_index))
s1, p1, o1 = new_triple[0].split('\t')
s2, p2, o2 = new_triple[1].split('\t')
tmp_knowledge_list.append([s1, p1, o1])
tmp_knowledge_list.append([s2, p2, o2])
result_json['knowledge'] = tmp_knowledge_list
del result_json['q_list'][0]
if result_json['q_list']:
system_output = '\n' + NLG([
result_json['q_list'][0][0] + '\t' +
result_json['q_list'][0][1] + '\t' +
result_json['q_list'][0][2]
], 'Knowledge_question')[0]
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
if result_json['knowledge']:
system_output = ''
for triple in result_json['knowledge']:
if 'flagship/dialogid' not in triple:
system_output += NLG(['\t'.join(triple)],
'Knowledge_inform')[0] + '\n'
userDB_json['command'] = 'REGISTER'
userDB_json['triple'] = result_json['knowledge']
Dialog_Manager.UserDBaccess(userDB_json)
result_json['entities'] = []
result_json['knowledge'] = []
system_output += '감사합니다'
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
result_json['entities'] = []
system_output = '감사합니다'
result_json['utterance'] = system_output
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
else:
print('input error')
result_json = input_json
result_json['utterance'] = 'input error'
else:
print('input error')
result_json = input_json
result_json['utterance'] = 'input error'
return jsonify(result_json)
print(system_output)
dialogDB_json['mode'] = 'add_data'
dialogDB_json['utterance'] = system_output
dialogDB_json['speaker'] = 'system'
dialog_index = DialogDBaccess(dialogDB_json)
'''
user_input = input()
if user_input == '로그아웃':
break
elif user_input == '끝':
stop = True
break
dialogDB_json['utterance'] = user_input
dialogDB_json['speaker'] = 'user'
dialog_index = DialogDBaccess(dialogDB_json)
NLU_response = NLU.Frame_Interpreter(user_input)
#frame기반
if NLU_response['frames']:
KB_language['frames'] = NLU_response['frames']
#for now_frame in KB_language['frames']:
now_frame = KB_language['frames'][-1]
question_list = []
with open('./frame_info_full.json', 'r',
encoding='utf-8') as f:
frame_json_data = json.load(f)
frame_data = frame_json_data[now_frame['frame']]
for ele in frame_data['arguments']:
if ele['coreType'] == 'Core':
question_list.append(ele['fe'])
print(question_list)
for exi_ele in now_frame['ele']: