1 # -*- coding: utf-8 -*-
 2 import psycopg2
 3 import math
 4 #读取数据库中用户打分数据存入musicdata
 5 conn = psycopg2.connect(database = ‘mytest‘, user = ‘postgres‘, password = ‘piao123‘, host = ‘localhost‘,port=‘5432‘)
 6 cur = conn.cursor()
 7 cur.execute("SELECT * FROM musicdata;")
 8 data=cur.fetchall()
 9 musicdata={user[0]:user[1:] for user in data}
10 cur.close()
11 conn.close()
12 print‘read database OK‘
13 
14 #将每首音乐的打分分别存为字典形式
15 musicScores=[]
16 n=len(musicdata)
17 m=len(musicdata[‘user1‘])
18 
19 #listenedsongs=set()
20 listened={}
21 unlistened={}
22 for i in range(m):
23     musicScores.append({})
24     for j in range(n):
25         if not unlistened.get(‘user‘+str(j+1)):
26             unlistened[‘user‘+str(j+1)]=[]
27         if not listened.get(‘user‘+str(j+1)):
28             listened[‘user‘+str(j+1)]=[]
29         if musicdata[‘user‘+str(j+1)][i]!=0:
30             musicScores[i][‘user‘+str(j+1)]=musicdata[‘user‘+str(j+1)][i]
31             #listenedsongs.add(‘song‘+str(i+1))
32             listened[‘user‘+str(j+1)].append(‘song‘+str(i+1))
33         else:
34             unlistened[‘user‘+str(j+1)].append(‘song‘+str(i+1))
35 musicdata=‘‘
36 print ‘musicScores is OK‘
37 #计算机song_i和song_j的相似度
38 sim={}
39 for i in range(m):
40     for j in range(m):
41         if j>i:
42             muiltSum=0
43             leftSquareSum=0
44             rightSquareSum=0
45             for user in range(n):
46                 if musicScores[i].get(‘user‘+str(user+1)) and musicScores[j].get(‘user‘+str(user+1)):
47                     muiltSum+=musicScores[i][‘user‘+str(user+1)]*musicScores[j][‘user‘+str(user+1)]
48                     leftSquareSum+=musicScores[i][‘user‘+str(user+1)]*musicScores[i][‘user‘+str(user+1)]
49                     rightSquareSum+=musicScores[j][‘user‘+str(user+1)]*musicScores[j][‘user‘+str(user+1)]
50             sim[(‘song‘+str(i+1),‘song‘+str(j+1))]=muiltSum/(math.sqrt(leftSquareSum)*math.sqrt(rightSquareSum))
51 
52 print ‘sim is OK‘
53 pred={}
54 for user in range(n):
55     SumSim=0
56     SumUsim=0
57     for song_p in unlistened[‘user‘+str(user+1)]:
58         for song_i in listened[‘user‘+str(user+1)]:
59             if song_p>song_i:
60                 SumSim += sim[(song_i,song_p)]
61                 SumUsim += musicScores[int(song_i[4:])-1][‘user‘+str(user+1)]*sim[(song_i,song_p)]
62             else:
63                 SumSim += sim[(song_p,song_i)]
64                 SumUsim += musicScores[int(song_i[4:])-1][‘user‘+str(user+1)]*sim[(song_p,song_i)]
65         if SumSim!=0:
66             pred[(user+1,song_p)]=SumUsim/SumSim
67 print ‘pred is OK‘
68 
69 
70 ‘‘‘
71 Created on 2014年4月21日
72 
73 @author: Administrator
74 ‘‘‘