def jsonToSQLJob():
print(">>>>>>>>>>Initiating Persistence to SQLLite>>>>>>>>>>")
conn = create_connection(get_database_path())
drop_table(conn, get_drop_query('D'))
create_table(conn, get_create_query('D'))
drop_table(conn, get_drop_query('C'))
create_table(conn, get_create_query('C'))
with open('participation.json') as json_file:
data = json.load(json_file)
i = 1
for p in data['all']:
insert_table(
conn, "INSERT INTO participation values(" + str(i) + "," +
str(p) + ");")
i = i + 1
with open('punch_card.json') as json_file:
data = json.load(json_file)
for p in data:
insert_table(
conn, "INSERT INTO punch_card values(" + str(p[0]) + "," +
str(p[1]) + "," + str(p[2]) + ");")
print("<<<<<<<<<<<<<<<<<<Persistence to SQLLite Complete<<<<<<<<<<<<<")
def SolutionToDbJob(type, solutionA, solutionB):
conn = create_connection(get_database_path())
if (type == "findDayOfMaximumCommit"):
insert_table(
conn, "INSERT INTO day_with_max_commit values(" + str(solutionA) +
",'" + str(solutionB) + "');")
if (type == "findTheWeekWithMaxCommit"):
insert_table(
conn,
"INSERT INTO week_with_max_commit values(" + str(solutionA) + ");")
def domain_decision(parameter_list):
global stopword_set
corrected_parameter_list = []
for parameter in parameter_list:
b = check_in_vocabulary(parameter)
log("check_in_vocabulary = " + str(b))
if not b:
words = jieba_cut_vocabulary(parameter)
corrected_parameter_list.extend(words)
else:
corrected_parameter_list.append(parameter)
topics_cost = get_topic_cost_by_words(corrected_parameter_list)
selected_topics = select_topic(topics_cost)
global parameter_similar_list
result_parameter_order_list = []
#排序parameter相似度-----
for topic in selected_topics:
index = topic_list_en.index(topic)
log(topic + ', index = ' + str(index))
object = {'topic': topic}
parameter_list = []
for i in range(len(parameter_similar_list)):
word = ''
obj = ()
for key in parameter_similar_list[i].keys():
word = key
obj = (word, parameter_similar_list[i][word][index])
#log('word = '+word+', score = '+str(parameter_similar_list[i][word][index]))
parameter_list.append(obj)
parameter_list = sorted(parameter_list,
key=lambda x: x[1],
reverse=True)
parameter_array = []
for tuple in parameter_list:
parameter_array.append(tuple[0])
object['parameter'] = parameter_array
result_parameter_order_list.append(object)
log('parameter_list = ' + str(result_parameter_order_list))
#排序parameter相似度-----end
# [{'topic':'music', 'parameter':['五月天','溫柔']}]
log("選中 ****** " + str(selected_topics) + " ******")
return [selected_topics, topics_cost, result_parameter_order_list]
def draw():
surface.fill((0, 0, 0))
particle_system.draw(surface)
player1.draw(surface)
for asteroid in asteroids:
asteroid.draw(surface)
#########
# surf_level = fonts[16].render("Level: "+str(level), True, (255,255,255))
# surface.blit(surf_level,(10,10))
#
# surf_lives = fonts[16].render("Lives: "+str(max([player1.lives,0])), True, (255,255,255))
# surface.blit(surf_lives,(10,25))
#
# surf_score = fonts[16].render("Score: "+str(player1.score), True, (255,255,255))
# surface.blit(surf_score,(screen_size[0]-surf_score.get_width()-10,10))
#
# surf_highscore = fonts[16].render("High Score: "+str(hs), True, (255,255,255))
# surface.blit(surf_highscore,(screen_size[0]-surf_highscore.get_width()-10,25))
if player1.lives >= 0:
surf_remain = fonts[16].render(
"Asteroids Left: " + str(len(asteroids)), True, (255, 255, 255))
surface.blit(surf_remain,
(10, screen_size[1] - surf_remain.get_height() - 10))
if level_text_brightness > 0.0:
col = rndint(255.0 * level_text_brightness)
surf_level = fonts[32].render("Level " + str(level), True,
(col, col, col), (0, 0, 0))
pos = [(screen_size[0] / 2.0) - (surf_level.get_width() / 2.0),
(screen_size[1] / 2.0) - (surf_level.get_height() / 2.0)]
surface.blit(surf_level, pos, special_flags=BLEND_MAX)
else:
surf_level = fonts[32].render("F2 Starts New Game", True,
(255, 255, 255), (0, 0, 0))
pos = [(screen_size[0] / 2.0) - (surf_level.get_width() / 2.0),
(screen_size[1] / 2.0) - (surf_level.get_height() / 2.0)]
surface.blit(surf_level, pos, special_flags=BLEND_MAX)
image_data = pygame.surfarray.array3d(pygame.display.get_surface())
pygame.display.flip()
return image_data
def domain_decision_detail(list):
result = ''
log('domain_decision_detail = ' + str(len(list)))
detail_msg = ''
domain_result = domain_decision(list)
if len(domain_result) == 0:
return "key word not in vocabulary"
result = result + str(list) + '<br>'
for i in range(TOPIC_SIZE):
detail_msg = detail_msg + topic_list[i] + ':' + str(
domain_result[1][i]) + '<br>'
result = result + str(domain_result[0]) + '<br>' + detail_msg
log('result = ' + str(result))
return json.dumps(result, ensure_ascii=False)
def get_topic_cost_by_words(words):
segment_cost_list = []
result = []
words_in_vocabulary = []
global parameter_similar_list
parameter_similar_list = []
for word in words:
topic_cost_by_word = get_topic_cost(word)
if len(topic_cost_by_word) == TOPIC_SIZE:
words_in_vocabulary.append(word)
segment_cost_list.append(topic_cost_by_word)
for i in range(len(words_in_vocabulary)):
object = {}
object[words_in_vocabulary[i]] = segment_cost_list[i]
parameter_similar_list.append(object)
log('[ ' + str(words_in_vocabulary[i]) + ' ] cost list')
msg = ''
for j in range(TOPIC_SIZE):
msg = msg + topic_list[j] + ':' + str(
segment_cost_list[i][j]) + " | "
log('>>> ' + msg)
if len(segment_cost_list) == 0:
return []
#[[ 0.11703352075249501, .....],[0.5000000000000002, .......],[0.5000000000000002, ......]]
# >>> [ 0.11703352075249501, 0.5000000000000002, .......,]
log(str(words))
log(' === weighting handling ===')
for i in range(TOPIC_SIZE):
cost = 0
words_total_len = 0
for j in range(len(segment_cost_list)):
cost = cost + segment_cost_list[j][i] * len(words[j]) #TODO
words_total_len = words_total_len + len(words[j])
if words_total_len > 0:
cost = cost / words_total_len
result.append(cost)
log(topic_list[i] + ' : ' + str(cost))
return result
def compute_accuracy(training_sentences, is_show):
total = 0
match_conuter = 0
selected_topic_counter = 0
for i in range(len(training_sentences)):
total = total + len(training_sentences[i])
for sentence in training_sentences[i]:
sentence_seg = sentence.split(",")
topic_list = demo_domain.domain_decision(sentence_seg)[0]
selected_topic_counter = selected_topic_counter + len(topic_list)
# log(str(sentence_seg)+' = '+str(topic_list), is_show)
if i==0 and demo_domain.topic_list_en[0] in topic_list:
match_conuter = match_conuter + 1
elif i==1 and demo_domain.topic_list_en[1] in topic_list:
match_conuter = match_conuter + 1
elif i==2 and demo_domain.topic_list_en[2] in topic_list:
match_conuter = match_conuter + 1
elif i==3 and demo_domain.topic_list_en[3] in topic_list:
match_conuter = match_conuter + 1
elif i==4 and demo_domain.topic_list_en[4] in topic_list:
match_conuter = match_conuter + 1
elif i==5 and demo_domain.topic_list_en[5] in topic_list:
match_conuter = match_conuter + 1
elif i==6 and demo_domain.topic_list_en[6] in topic_list:
match_conuter = match_conuter + 1
elif i==7 and demo_domain.topic_list_en[7] in topic_list:
match_conuter = match_conuter + 1
elif i==8 and demo_domain.topic_list_en[8] in topic_list:
match_conuter = match_conuter + 1
else:
log('not match sentence = '+str(sentence_seg)+' = '+str(topic_list), is_show)
cover_rate = round(match_conuter*100/total,2)
if selected_topic_counter == 0:
accuracy_rate = 0
else:
accuracy_rate = round(match_conuter*100/selected_topic_counter,2)
log('Cover rate = '+str(cover_rate)+'%', is_show)
log('Accuracy rate = '+str(accuracy_rate)+'%', is_show)
return cover_rate, accuracy_rate
def start_training(training_sentences):
#init
demo_domain.offset = 0.01
learning = 0.002
max_integrating_rate = [0,0]
max_cover_rate = [0,0]
max_accuracy_rate = [0,0]
final_offset ={}
lower_bound = 2*1/demo_domain.TOPIC_SIZE*100
while(True):
rate = compute_accuracy(training_sentences, False)
if integrating_cal(rate[0], rate[1]) > integrating_cal(max_integrating_rate[0], max_integrating_rate[1]):
max_integrating_rate = rate
final_offset['integrating'] = demo_domain.offset
print('[update] integrating offset = '+str(final_offset['integrating'])+', CoverRate = '+str(max_integrating_rate[0])+'%, AccuracyRate = '+str(max_integrating_rate[1])+'%')
if rate[0] > max_cover_rate[0]:
max_cover_rate = rate
final_offset['cover'] = demo_domain.offset
print('[update] cover offset = '+str(final_offset['cover'])+', CoverRate = '+str(max_cover_rate[0])+'%, AccuracyRate = '+str(max_cover_rate[1])+'%')
if rate[1] > max_accuracy_rate[1]:
max_accuracy_rate = rate
final_offset['accuracy'] = demo_domain.offset
print('[update] accuracy offset = '+str(final_offset['accuracy'])+', CoverRate = '+str(max_accuracy_rate[0])+'%, AccuracyRate = '+str(max_accuracy_rate[1])+'%')
if rate[0] == 100 and rate[1] < lower_bound:
break
demo_domain.offset = round(demo_domain.offset + learning, 3)
print('='*80)
print('[final] integrating offset = '+str(final_offset['integrating'])+', CoverRate = '+str(max_integrating_rate[0])+'%, AccuracyRate = '+str(max_integrating_rate[1])+'%')
print('[final] cover offset = '+str(final_offset['cover'])+', CoverRate = '+str(max_cover_rate[0])+'%, AccuracyRate = '+str(max_cover_rate[1])+'%')
print('[final] accuracy offset = '+str(final_offset['accuracy'])+', CoverRate = '+str(max_accuracy_rate[0])+'%, AccuracyRate = '+str(max_accuracy_rate[1])+'%')
demo_domain.offset = final_offset['integrating']
compute_accuracy(training_sentences, True)
def get_topic_cost(key_word):
topic_cost_list = []
global topic_list
for i in range(len(topic_list)):
max_cost = -1
feature_cost_list = []
#選出最大feature當作topic cost
for topic in topic_feature_list[i]:
try:
res = model.similarity(key_word, topic)
if (res > max_cost):
max_cost = res
feature_cost = {}
feature_cost[topic] = res
feature_cost_list.append(feature_cost)
except Exception as e:
log('get_topic_cost : ' + repr(e))
return []
log(key_word + ' >>> feature list : ' + str(feature_cost_list))
# topic_cost_object['domain'] = feature_cost_list
topic_cost_list.append(max_cost)
log(' --------------------------------------------- ')
return topic_cost_list
try:
model.most_similar(word, topn=1)
return True
except Exception as e:
return False
def log(str):
if is_log_showing:
print(str)
if __name__ == '__main__':
init_model()
while True:
try:
query = input()
q_list = query.split()
if len(q_list) == 1:
result = domain_decision(q_list[0])
domain_decision(q_list[0])
print("相似詞前 100 排序")
res = model.most_similar(q_list[0], topn=100)
for item in res:
print(item[0] + "," + str(item[1]))
except Exception as e:
print(repr(e))
def write_hs():
f = open("hs.txt", "wb")
f.write(str(hs).encode())
f.close()