cikm12-vs-cf-sourcecode/weighted_KendallTauRank_specialty_degree_sim.py at master · zhenv5/cikm12-vs-cf-sourcecode · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
'''
vector space model for computing similarity
Author: Sun Jiankai
'''
import datetime
import pprocess
import math
from meta_weight import item_num
from meta_weight import user_num
from meta_weight import limit
from meta_weight import score_weight_log
from meta_weight import confidence_weight_lambda_log
from meta_weight import confidence_weight_a
from meta_method_weight import getSequence
#trainfilename='dataset/split_train_1m.data'
trainfilename='dataset/split_eachmovie_train.data'
trainfile=open(trainfilename,'r')
user_itempairs_tf_idf_list=[]
user_itempairs_idf_list=[[0 for i in range(item_num)] for j in range(item_num)]
user_itempairs_tf_list=[]
print 'for test'
def calculate_tf_idf(t):
    global user_itempairs_idf_list
    user_id,item_pairs_list=t
    user_itempairs_tf_idf_list_temp=[]
    for j in range(len(item_pairs_list)):
        item_i,item_j,tf,score=item_pairs_list[j]
        count_positive=user_itempairs_idf_list[int(item_i)-1][int(item_j)-1]
        count_negative=user_itempairs_idf_list[int(item_j)-1][int(item_i)-1]
        indicator=0
        if count_negative!=0 and count_negative!=0:
            if score>0:
                weight_lambda=round(math.log(1+count_negative*1.0/count_positive,confidence_weight_lambda_log),8)
                indicator=1
            else:
                weight_lambda=round(math.log(1+count_positive*1.0/count_negative,confidence_weight_lambda_log),8)
                indicator=-1
        else:
            weight_lambda=0
        weight_alpa=round(math.log(1.00+(confidence_weight_a-1)*(count_positive+count_negative)/user_num,confidence_weight_a),8)
        idf=round(weight_alpa*weight_lambda+1.00-weight_alpa,8)
        tf_idf=round(tf*idf,8)
        user_itempairs_tf_idf_list_temp.append((item_i,item_j,tf_idf,indicator))
    print 'user_id '+user_id+' is ok,tf-idf'
    return i,user_id,user_itempairs_tf_idf_list_temp
def calculate_sim(t):
    i,user_id_i,user_itempairs_tf_idf_list_temp_i=t
    global user_itempairs_tf_idf_list
    user_user_sim_list=[]
    for j in range(i+1,len(user_itempairs_tf_idf_list)):
        j,user_id_j,user_itempairs_tf_idf_list_temp_j=user_itempairs_tf_idf_list[j]
        len_i=len(user_itempairs_tf_idf_list_temp_i)
        len_j=len(user_itempairs_tf_idf_list_temp_j)
        pointer_i=0
        pointer_j=0
        euclidean=0
        inner_sum=0

        while pointer_i<len_i and pointer_j<len_j:
            item_i_i=int(user_itempairs_tf_idf_list_temp_i[pointer_i][0])
            item_j_i=int(user_itempairs_tf_idf_list_temp_i[pointer_i][1])
            item_i_j=int(user_itempairs_tf_idf_list_temp_j[pointer_j][0])
            item_j_j=int(user_itempairs_tf_idf_list_temp_j[pointer_j][1])
            if item_i_i==item_i_j:
                if item_j_i==item_j_j:
                    tf_idf_i=float(user_itempairs_tf_idf_list_temp_i[pointer_i][2])
                    tf_idf_j=float(user_itempairs_tf_idf_list_temp_j[pointer_j][2])
                    indicator_i=int(user_itempairs_tf_idf_list_temp_i[pointer_i][3])
                    indicator_j=int(user_itempairs_tf_idf_list_temp_j[pointer_j][3])
                    euclidean+=round(tf_idf_i*tf_idf_j,8)
                    inner_sum+=round(tf_idf_i*tf_idf_j*indicator_i*indicator_j,8)


                    pointer_i+=1
                    pointer_j+=1
                elif item_j_i<item_j_j:
                    pointer_i+=1
                else:
                    pointer_j+=1
            elif item_i_i<item_i_j:
                pointer_i+=1
            else:
                pointer_j+=1
        if euclidean!=0 :
            sim=round(inner_sum/euclidean,8)
        else:
            sim=0
        print 'i j sim is:'+user_id_i+' '+user_id_j+' '+str(sim)
        user_user_sim_list.append((user_id_i,user_id_j,sim))
    return user_user_sim_list

def calculate_tf(t):
    user_id,templist=t
    item_nums=len(templist)
    item_pairs_list=[]
    for i in range(item_nums-1):
        item_i,score_i=templist[i]
        for j in range(i+1,item_nums):
            item_j,score_j=templist[j]
            #append to item pairs list
            if int(score_i)!=int(score_j):
                #item_pairs_list.append((item_i,item_j,int(score_i)-int(score_j)))
                tf=round(math.log(1+math.fabs((int(score_i)-int(score_j))),score_weight_log),8)
                item_pairs_list.append((item_i,item_j,tf,int(score_i)-int(score_j)))
    print 'tf userid:'+user_id
    return user_id,item_pairs_list
def main(resultfilename):
    start_1=datetime.datetime.now()
    print str(start_1)
    resultfile=open(resultfilename,'w')
    #used for computing tf
    sequence_tf=getSequence(trainfile)
    #used for computing idf
    global user_itempairs_idf_list,user_itempairs_tf_list
    user_user_sim_matrix=[[0 for i in range(user_num)]for j in range(user_num)]
    start_2=datetime.datetime.now()
    print str(start_2)+' 2 used '+str(start_2-start_1)+'total used '+str(start_2-start_1)
    results=pprocess.pmap(calculate_tf,sequence_tf,limit)
    for result in results:
        user_id,item_pairs_list=result
        user_itempairs_tf_list.append((user_id,item_pairs_list))
        for i in range(len(item_pairs_list)):
            item_i,item_j,tf,score=item_pairs_list[i]
            if score>0:
                user_itempairs_idf_list[int(item_i)-1][int(item_j)-1]+=1
            else:
                user_itempairs_idf_list[int(item_j)-1][int(item_i)-1]+=1
        #print 'user_id for tf '+user_id
    start_3=datetime.datetime.now()
    print str(start_3)+' 3 used '+str(start_3-start_2)+'total used '+str(start_3-start_1)
    global user_itempairs_tf_idf_list
    results=pprocess.pmap(calculate_tf_idf,user_itempairs_tf_list,limit)
    for result in results:
        user_itempairs_tf_idf_list.append(result)
    del user_itempairs_tf_list
    del user_itempairs_idf_list
    start_4=datetime.datetime.now()
    print str(start_4)+' 4 used '+str(start_4-start_3)+'total used '+str(start_4-start_1)
    results=pprocess.pmap(calculate_sim,user_itempairs_tf_idf_list,limit)
    for result in results:
        for i in range(len(result)):
            u_i=int(result[i][0])-1
            u_j=int(result[i][1])-1
            sim=float(result[i][2])
            user_user_sim_matrix[u_i][u_j]=sim
        #print str(u_i)+' is computed ok'
    start_5=datetime.datetime.now()
    del user_itempairs_tf_idf_list
    print str(start_5)+' 5 used '+str(start_5-start_4)+'total used '+str(start_5-start_1)
    for i in range(user_num):
        for j in range(user_num):
            resultfile.write(str(user_user_sim_matrix[i][j]))
            resultfile.write(' ')
        resultfile.write('\n')
    start_6=datetime.datetime.now()
    print str(start_6)+' 6 used '+str(start_6-start_5)+'total used '+str(start_6-start_1)
    resultfile.close()
    trainfile.close()
    start_7=datetime.datetime.now()
    print str(start_7)+' 7 used '+str(start_7-start_6)+'total used '+str(start_7-start_1)
    print 'sim end'
    print '********************************'

print 'start'
print str(user_num)
main('afterdone/user_user_similarity/user_user_sim_eachmovie_weight.data')