def __init__(self, threadMaxNums=1):
self.THREAD_MAX = threadMaxNums
self.currentAvailThreads = self.THREAD_MAX
self.availThreadCondi = threading.Condition()
self.databaseWriteInCondi = threading.Condition()
self.lock_availableThreads = threading.Lock()
self.dataSpace = ThreadSafeData()
#initialize the user seed list
self.dataSpace.userSeedsList.put(340056317)
self.dataSpace.userSeedsList.put(350714427)
self.dataSpace.userSeedsList.put(20888663)
self.dataSpace.userSeedsList.put(588707084)
db_temp = db_cls("pj_data.db")
user_tables = db_temp.read_Data("Select * from Follow")
for i in range(3):
self.dataSpace.userSeedsList.put(
list(random.choice(user_tables))[1])
db_temp.close_db()
candidate_songs_file = open("./candidate_songs.txt")
lines = candidate_songs_file.readlines()
for line in lines:
seed = int(line.strip())
# debug_print_thread(seed, True)
self.dataSpace.songSeedsList.put(seed)
self.__first_db_initialize_flag = False
def main():
#从数据库中读取所需信息,并做格式化预处理
db = db_cls(db_filename="pj_data.db")
user_list = db.read_Data(sql_query="Select id FROM User_Table")
user_list = [ite[0] for ite in user_list]
#item_list_dict = dict(db.read_Data(sql_query="Select id,song_name FROM Song"))
#item_list = list(item_list_dict)
item_list1 = db.read_Data(sql_query="Select song_id FROM User2Song")
item_list2 = db.read_Data(sql_query="Select song_id FROM SongList2Song")
item_list = list(set(item_list1 + item_list2))
item_list = [ite[0] for ite in item_list]
user_list = sorted(user_list)
item_list = sorted(item_list)
item_list_dict = dict(
db.read_Data(sql_query="Select id,song_name FROM Song"))
for item in item_list:
item_list_dict[item] = item_list_dict.get(item, "歌名未知")
#创建“最近听歌”字典,key为user_id,value为该用户最近听歌的歌曲id组成的列表
recent_listen = db.read_Data(sql_query="Select * FROM User2Song")
recent_listen_dict = dict()
for it in recent_listen:
recent_listen_dict[it[0]] = recent_listen_dict.get(it[0], []) + [it[1]]
for it in user_list:
recent_listen_dict[it] = recent_listen_dict.get(it, [])
#创建“歌单音乐”字典,key为user_id,value为该用户收藏歌单中歌曲id组成的列表
list_music = db.read_Data(sql_query="Select a.user_id,b.song_id \
FROM User2SongList a INNER JOIN SongList2Song b\
ON a.songlist_id = b.songlist_id")
list_music_dict = dict()
for it in list_music:
list_music_dict[it[0]] = list_music_dict.get(it[0], []) + [it[1]]
for it in user_list:
list_music_dict[it] = list_music_dict.get(it, [])
#获取用户-音乐评价矩阵
user_item_matrix = cf.get_matrix(user_list, item_list, recent_listen_dict,
list_music_dict)
#获取用户相似度矩阵
user_similarity_matrix = cf.get_similarity(user_item_matrix)
#获取基于用户相似性得到的推荐矩阵
Recommender = cf.get_recommender(user_item_matrix, user_similarity_matrix,
5)
#为id为393718733的用户推荐5首歌曲
#cf.recommend(393718733, user_list, item_list_dict, Recommender, 20)
cf.recommend(2141581764, user_list, item_list_dict, Recommender, 10)
def create_db(self, db_filename="pj_data.db"):
self.db = db_cls(db_filename=db_filename)
global visited_song_list
global visited_user_list
vsongs = self.db.read_Data("Select * from Song")
for eachSong in vsongs:
id = list(eachSong)[0]
if id not in visited_song_list:
visited_song_list.append(id)
vusers = self.db.read_Data("Select * from User_Table")
for eachUser in vusers:
id = list(eachUser)[0]
if id not in visited_user_list:
visited_user_list.append(list(eachUser)[0])
def user_song_relation(db_filename):
# 载入数据库
db = db_cls(db_filename)
# 第一次使用可创建索引，加快数据库查询速度
# db.create_index()
# 分别载入歌曲与用户的数据
song_pos2id, song_id2pos, song_num = song2user_load_songs(db)
user_pos2id, user_id2pos, user_num = song2user_load_users(db)
# 构建歌曲-用户关系邻接表
R_song = [[] for i in range(song_num)]
R_user = [[] for i in range(user_num)]
song_user_relation = db.read_Data(sql_query="Select * from User2Song")
for user_id, song_id in song_user_relation:
user_pos = user_id2pos[user_id]
song_pos = song_id2pos[song_id]
R_song[song_pos].append(user_pos)
R_user[user_pos].append(song_pos)
db.close_db()
return song_pos2id, song_id2pos, song_num, user_pos2id, user_id2pos, user_num, R_song, R_user
start_time = time.time()
[song_infos_list, song2singer_list] = ScrapeAndSaveSong(seedUrl, maxNum)
# db.write_Song_infos(song_infos_list, song2singer_list)
global_song_infos_list.extend(song_infos_list)
global_song2singer_list.extend(song2singer_list)
end_time = time.time()
print('process finished with execution time ', end_time - start_time)
global global_count
global_count += 1
# db.create_index()
if __name__ == "__main__":
db = db_cls(db_filename="pj_data.db")
thread1 = threading.Thread(
target=Thread_Scrapy,
args=('https://music.163.com/#/song?id=31877628', 25))
thread2 = threading.Thread(target=Thread_Scrapy,
args=('https://music.163.com/#/song?id=186453',
25))
thread3 = threading.Thread(
target=Thread_Scrapy,
args=('https://music.163.com/#/song?id=399410693', 25))
thread4 = threading.Thread(target=Thread_Scrapy,
args=('https://music.163.com/#/song?id=1989355',
25))
thread1.start()
thread2.start()
c = 0.8
# 迭代次数
iter_num = 20
# 作为样本的歌曲，用于观察迭代后得到的相似歌曲结果
sample_songs = [
29947420, 461347998, 31356499, 1313354324, 436514312, 483671599
] # 用于我们自己爬的数据
#sample_songs = [139764, 144619, 408055928, 180280] # 用于老师给的数据
if __name__ == "__main__":
# 从数据库中载入数据
song_pos2id, song_id2pos, song_num, user_pos2id, user_id2pos, user_num, R_song, R_user = user_song_relation(
db_filename)
db = db_cls(db_filename)
# 分别构建歌曲及用户相似度矩阵，初始为单位阵
S_song = np.eye(song_num)
S_user = np.eye(user_num)
# Simrank 迭代
for iter in range(iter_num):
print("Iteration: " + str(iter + 1))
# 遍历歌曲相似矩阵，通过用户相似度来计算歌曲相似度
for x in tqdm(range(song_num), unit='rows'):
for y in range(song_num):
if x < y:
S_val = 0
# 分别连接歌曲x和y的用户列表