-
Notifications
You must be signed in to change notification settings - Fork 3
/
choose_sample_new.py
257 lines (246 loc) · 11.2 KB
/
choose_sample_new.py
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
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
import os
import sys
import random
import numpy as np
import pandas as pd
import xlrd
from pandas import Series, DataFrame
def Test():
# 超参数设置
r1 = 0.5
train = {}
test = {}
cc = 0;
ccc = 0;
flag = 1
count = 0
df_count = 0
# 初始化存储array
# 初始化array为-1，列数为100
store = np.arange(1000)
ii = 0
n = 0
for iii in store:
store[ii] = -1
ii += 1
# start 计算函数的数据
# ent代表实体，rel代表关系
ent = np.zeros([40943, 100])
rel = np.zeros([18, 100])
e1n = np.ones([1, 100])
en1 = np.ones([100, 1])
valid = []
n = 0
ent_i = 0
train2id = []
# end
# 读取所有的embedding并将它放入数组中
with open('/Users/ihao/Desktop/entity2vec_wn18.txt') as f:
nn = 0
for line in f:
n = 0
nn += 1
# =========此处要修改n的值为列数=========
while n < 100:
ent[ent_i][n] = line.strip().split('\t')[n]
n += 1
ent_i += 1
print('============')
print(nn)
iii = 0
# / Users / ihao / Desktop / embeded / rel_new.txt
with open('/Users/ihao/Desktop/relation2vec_wn18.txt') as f2:
for line in f2:
n = 0
# =========此处要修改n的值为列数=========
while n < 100:
rel[iii][n] = line.strip().split('\t')[n]
n += 1
iii += 1
split_n = 0
split_n1 = 0
split_n2 = 0
out = []
nor = []
df_n = 0
n = 0
# df = pd.read_excel('/Users/ihao/Desktop/fb15k237Store.xlsx', sheet_name='Sheet1', na_values='n/a')
# df_t = pd.read_excel('/Users/ihao/Desktop/fb15k237Store.xlsx', sheet_name='test', na_values='n/a')
fenmu = 0
# 为了求afax的分母
# =========此处要修改路径=========
# with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/FB15K237/test2id.txt') as test2id9:
with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/WN18/test2id.txt') as test2id9:
# 直接遍历test2id中的前100个三元组
for test_line in test2id9:
flag = 1
ccc = 0
# 忽略掉第一行的行数值
if cc == 0:
cc += 1
continue
# 如果到了100个，就退出
if count == 99:
break
dd = 0
# 进入train2id中查找
# =========此处要修改路径=========
# with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/FB15K237/train2id.txt') as t2id9, open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/FB15K237/train2id.txt') as t2id29:
with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/WN18/train2id.txt') as t2id9, open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/WN18/train2id.txt') as t2id29:
for train_line in t2id9:
if dd == 0:
dd += 1
continue
# 如果相等，说明在 train2id 中的值在 test2id 中出现过
if (test_line.split(' ')[0:]) == (train_line.split(' ')[0:]):
flag = 0
continue
# 说明test在train中还没有出现过
if flag == 1:
head_in_100 = int(test_line.split(' ')[0])
# 判断是否出现过 head
if (store == head_in_100).any():
# 出现过则直接跳过
continue
else:
# 未出现，则进行下一步操作
df_t_n = 0
iiii = 0
for check_line in t2id29:
# 跳过首行的行数值
if df_t_n == 0:
df_t_n += 1
continue
# icount = 0
# 在train2id中查找head相同的值
if int(check_line.split(' ')[0]) == head_in_100:
df_inh = int(check_line.split(' ')[0])
df_int = int(check_line.strip().split(' ')[1])
df_inr = int(check_line.strip().split(' ')[2])
# 求出 (h1*r1-t1)^2 的值，相加，最后开平方
fenmu += np.power(ent[df_inh] * rel[df_inr] - ent[df_int], 2)
print(fenmu)
count += 1
# 对分母开根号，得到想要的值
fenmu = np.sqrt(fenmu)
# 将计数器还原
count = 0
cc = 0
# =========此处要修改路径=========
# with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/FB15K237/test2id.txt') as test2id, open(
# '/Users/ihao/Desktop/sample_out.txt', 'w+') as outp:
with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/WN18/test2id.txt') as test2id, open(
'/Users/ihao/Desktop/sample_out_wn18.txt', 'w+') as outp:
# 直接遍历test2id中的前100个三元组
for test_line in test2id:
flag = 1
ccc = 0
# 忽略掉第一行的行数值
if cc == 0:
cc += 1
continue
# 如果到了100个，就退出
if count == 99:
break
dd = 0
# 进入train2id中查找
# =========此处要修改路径=========
# with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/FB15K237/train2id.txt') as t2id, open(
# '/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/FB15K237/train2id.txt') as t2id2:
with open('/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/WN18/train2id.txt') as t2id, open(
'/Users/ihao/Desktop/OpenKE-OpenKE-PyTorch/benchmarks/WN18/train2id.txt') as t2id2:
for train_line in t2id:
if dd == 0:
dd += 1
continue
# 如果相等，说明在 train2id 中的值在 test2id 中出现过
if (test_line.split(' ')[0:]) == (train_line.split(' ')[0:]):
flag = 0
continue
# 说明test在train中还没有出现过
if flag == 1:
# hrt = str(test_line.split(' ')[0]) + ' ' + str(test_line.strip().split(' ')[1]) + ' ' + str(
# test_line.strip().split(' ')[2])
head_in_100 = int(test_line.split(' ')[0])
# print((a == 999).any())
# 判断是否出现过 head
if (store == head_in_100).any():
# 出现过则直接跳过
continue
else:
# 未出现，则进行下一步操作
df_t_n = 0
iiii = 0
for check_line in t2id2:
# 跳过首行的行数值
if df_t_n == 0:
df_t_n += 1
continue
# icount = 0
# 在train2id中查找head相同的值
if int(check_line.split(' ')[0]) == head_in_100:
df_inh = int(check_line.split(' ')[0])
df_int = int(check_line.strip().split(' ')[1])
df_inr = int(check_line.strip().split(' ')[2])
# 使用excel作为容器，但是考虑后还是np.array好
# df_t['ht'][icount] = check_line.split(' ')[0]
# df_t['tt'][icount] = check_line.strip().split(' ')[1]
# df_t['rt'][icount] = check_line.strip().split(' ')[2]
# df_t['point'][0] = calculate(df_inh, df_int, df_inr)
# 查找对应的 embedding 值
# f = ent[df_inh] + rel[df_inr] - ent[df_int]
# (hi*ri-ti)*ri
f = (ent[df_inh] * rel[df_inr] - ent[df_int]) * rel[df_inr]
f = np.array(f)
f = f.reshape(1, -1)
nor_x = np.linalg.norm(f)
# nor_x = np.sqrt(f.dot(f.T))
# f=||h+r-t|| ==> nor_x
# print(nor_x)
# hrt = str(nor_x) + "\n"
# 将nor_x 写入文件
# validout.write(hrt)
# 对 f 求偏导
ratio = f
# ratio1 = np.sqrt(f.dot(f.T))
afax = ratio / fenmu
# print(afax)
f2 = (ent[df_inh] + afax) + rel[df_inr] - ent[df_int]
f2 = np.array(f2)
f2 = f2.reshape(1, -1)
nor_x2 = np.linalg.norm(f2)
# nor_x2 = np.sqrt(f2.dot(f2.T))
n_hrt = float(-(nor_x - r1 * nor_x2))
# df_t['point'][icount] = n_hrt
#
# 改进--使用array
# 将三元组都存储到np.array中
if iiii == 0:
arr = np.array([[df_inh, df_int, df_inr, n_hrt]])
iiii += 1
else:
arr = np.append(arr, [[df_inh, df_int, df_inr, n_hrt]], axis=0)
# icount += 1
# df_t.sort_values('point', ascending=False)
# 将元素降序排列
if len(arr) != 0:
sort_arr = arr[arr[:, 3].argsort()[::-1]]
lin_n = 0
for lin in sort_arr:
# 直接忽略前面5个，相当于删除了
if lin_n < 5:
lin_n += 1
continue
# 写入到 txt
str_out = str(int(lin[0])) + ' ' + str(int(lin[1])) + ' ' + str(int(lin[2])) + ' ' + str(
lin[3]) + '\n'
print(str_out)
outp.write(str_out)
arr = ''
sort_arr = ''
store[n] = head_in_100
n += 1
print(n)
count += 1
if __name__ == '__main__':
Test()