-
Notifications
You must be signed in to change notification settings - Fork 1
/
data.py
690 lines (600 loc) · 23.8 KB
/
data.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
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
import torch
import random
import json
import numpy as np
import pdb
import torch.distributed as dist
import os
import os.path as osp
from collections import Counter
import pickle
import torch.nn.functional as F
from transformers import BertTokenizer
import torch.distributed
from tqdm import tqdm
import re
from utils import get_topk_indices, get_adjr
class EADataset(torch.utils.data.Dataset):
def __init__(self, data):
self.data = data
def __len__(self):
return len(self.data)
def __getitem__(self, index):
return self.data[index]
class Collator_base(object):
def __init__(self, args):
self.args = args
def __call__(self, batch):
# pdb.set_trace()
return np.array(batch)
# def load_data(logger, args):
# assert args.data_choice in ["DWY", "DBP15K", "FBYG15K", "FBDB15K"]
# if args.data_choice in ["DWY", "DBP15K", "FBYG15K", "FBDB15K"]:
# KGs, non_train, train_ill, test_ill, eval_ill, test_ill_ = load_eva_data(logger, args)
#
# elif args.data_choice in ["FBYG15K_attr", "FBDB15K_attr"]:
# pass
#
# return KGs, non_train, train_ill, test_ill, eval_ill, test_ill_
#
def load_eva_data(args):
if "OEA" in args.data_choice:
file_dir = osp.join(args.data_path, "OpenEA", args.data_choice)
else:
file_dir = osp.join(args.data_path, args.data_choice, args.data_split)
lang_list = [1, 2]
ent2id_dict, ills, triples, r_hs, r_ts, ids = read_raw_data(file_dir, lang_list)
e1 = os.path.join(file_dir, 'ent_ids_1')
e2 = os.path.join(file_dir, 'ent_ids_2')
left_ents,left_id2name = get_ids(e1,file_dir)
right_ents,right_id2name = get_ids(e2,file_dir)
id2name = {**left_id2name, **right_id2name}
if not args.data_choice == "DBP15K" and not args.data_choice == "OpenEA":
id2rel = get_id2rel(os.path.join(file_dir, 'id2relation.txt'))
elif args.data_choice == "OpenEA":
id2rel = get_id2rel(os.path.join(file_dir, 'rel_ids'))
else:
id2rel = None
ENT_NUM = len(ent2id_dict)
REL_NUM = len(r_hs)
np.random.shuffle(ills)
if args.mm:
if args.data_choice == "OpenEA":
img_vec_path = osp.join(args.data_path, f"OpenEA/pkl/{args.data_split}_id_img_feature_dict.pkl")
elif "FB" in file_dir:
img_vec_path = osp.join(args.data_path, f"pkls/{args.data_choice}_id_img_feature_dict.pkl")
else:
# fr_en
split = file_dir.split("/")[-1]
img_vec_path = osp.join(args.data_path, "pkls", args.data_split + "_GA_id_img_feature_dict.pkl")
assert osp.exists(img_vec_path)
img_features = load_img(ENT_NUM, img_vec_path)
print(f"image feature shape:{img_features.shape}")
if args.word_embedding == "glove":
word2vec_path = os.path.join(args.data_path, "embedding", "glove.6B.300d.txt")
elif args.word_embedding == 'bert':
pass
else:
raise Exception("error word embedding")
else:
img_features = None
name_features = None
char_features = None
# if args.data_choice == "DBP15K" and (args.w_name or args.w_char):
# assert osp.exists(word2vec_path)
# ent_vec, char_features = load_word_char_features(ENT_NUM, word2vec_path, args)
# name_features = F.normalize(torch.Tensor(ent_vec))
# char_features = F.normalize(torch.Tensor(char_features))
# print(f"name feature shape:{name_features.shape}")
# print(f"char feature shape:{char_features.shape}")
img_ill = None
if args.mm:
input_features = F.normalize(torch.Tensor(img_features))
img_ill = visual_pivot_induction(args, left_ents, right_ents, input_features, ills)
train_ill = np.array(ills[:int(len(ills) // 1 * args.data_rate)], dtype=np.int32)
test_ill_ = ills[int(len(ills) // 1 * args.data_rate):]
test_ill = np.array(test_ill_, dtype=np.int32)
test_left = torch.LongTensor(test_ill[:, 0].squeeze())
test_right = torch.LongTensor(test_ill[:, 1].squeeze())
left_non_train = list(set(left_ents) - set(train_ill[:, 0].tolist()))
right_non_train = list(set(right_ents) - set(train_ill[:, 1].tolist()))
print(f"#left entity : {len(left_ents)}, #right entity: {len(right_ents)}")
print(f"#left entity not in train set: {len(left_non_train)}, #right entity not in train set: {len(right_non_train)}")
rel_features = load_relation(ENT_NUM, triples, 1000)
print(f"relation feature shape:{rel_features.shape}")
if 'OpenEA' in args.data_choice:
a1 = os.path.join(file_dir, f'attr_triples_1')
a2 = os.path.join(file_dir, f'attr_triples_2')
att_features, num_att_left, num_att_right = load_attr_withNums(['oea', 'oea'], [a1, a2], ent2id_dict, file_dir,
topk=args.topk)
elif 'FB' in args.data_choice:
a1 = os.path.join(file_dir, 'FB15K_NumericalTriples.txt')
a2 = os.path.join(file_dir, 'DB15K_NumericalTriples.txt') if 'DB' in args.data_choice else os.path.join(file_dir, 'YAGO15K_NumericalTriples.txt')
att_features, num_att_left, num_att_right = load_attr_withNums(['FB15K','DB15K'] if 'DB' in args.data_choice else ['FB15K','YAGO15K'],[a1, a2], ent2id_dict, file_dir, topk=0)
else:
att1,att2 = args.data_split.split('_')
a1 = os.path.join(file_dir, f'{att1}_att_triples')
a2 = os.path.join(file_dir, f'{att2}_att_triples')
att_features, num_att_left, num_att_right = load_attr_withNums([att1,att2],[a1, a2], ent2id_dict, file_dir, topk=args.topk)
print(f"attribute feature shape:{len(att_features)}")
print("-----dataset summary-----")
print(f"dataset:\t\t {file_dir}")
print(f"triple num:\t {len(triples)}")
print(f"entity num:\t {ENT_NUM}")
print(f"relation num:\t {REL_NUM}")
print(f"train ill num:\t {train_ill.shape[0]} \t test ill num:\t {test_ill.shape[0]}")
print("-------------------------")
eval_ill = None
input_idx = torch.LongTensor(np.arange(ENT_NUM))
# pdb.set_trace()
# train_ill = EADataset(train_ill)
# test_ill = EADataset(test_ill)
return {
'ent_num': ENT_NUM,
'rel_num': REL_NUM,
'images_list': img_features,
'rel_features': rel_features,
'att_features': att_features,
'num_att_left': num_att_left,
'num_att_right': num_att_right,
'name_features': name_features,
'char_features': char_features,
'input_idx': input_idx,
'triples': triples,
'id2name':id2name,
'id2rel':id2rel,
'img_ill':img_ill
}, {"left": left_non_train, "right": right_non_train},left_ents,right_ents, train_ill, test_ill, eval_ill, test_ill_
def load_word2vec(path, dim=300):
"""
glove or fasttext embedding
"""
# print('\n', path)
word2vec = dict()
err_num = 0
err_list = []
with open(path, 'r', encoding='utf-8') as file:
for line in tqdm(file.readlines(), desc="load word embedding"):
line = line.strip('\n').split(' ')
if len(line) != dim + 1:
continue
try:
v = np.array(list(map(float, line[1:])), dtype=np.float64)
word2vec[line[0].lower()] = v
except:
err_num += 1
err_list.append(line[0])
continue
file.close()
print("err list ", err_list)
print("err num ", err_num)
return word2vec
def load_char_bigram(path):
"""
character bigrams of translated entity names
"""
# load the translated entity names
ent_names = json.load(open(path, "r"))
# generate the bigram dictionary
char2id = {}
count = 0
for _, name in ent_names:
for word in name:
word = word.lower()
for idx in range(len(word) - 1):
if word[idx:idx + 2] not in char2id:
char2id[word[idx:idx + 2]] = count
count += 1
return ent_names, char2id
def load_word_char_features(node_size, word2vec_path, args):
"""
node_size : ent num
"""
name_path = os.path.join(args.data_path, "DBP15K", "translated_ent_name", "dbp_" + args.data_split + ".json")
assert osp.exists(name_path)
save_path_name = os.path.join(args.data_path, "embedding", f"dbp_{args.data_split}_name.pkl")
save_path_char = os.path.join(args.data_path, "embedding", f"dbp_{args.data_split}_char.pkl")
if osp.exists(save_path_name) and osp.exists(save_path_char):
print(f"load entity name emb from {save_path_name} ... ")
ent_vec = pickle.load(open(save_path_name, "rb"))
print(f"load entity char emb from {save_path_char} ... ")
char_vec = pickle.load(open(save_path_char, "rb"))
return ent_vec, char_vec
word_vecs = load_word2vec(word2vec_path)
ent_names, char2id = load_char_bigram(name_path)
# generate the word-level features and char-level features
ent_vec = np.zeros((node_size, 300))
char_vec = np.zeros((node_size, len(char2id)))
for i, name in ent_names:
k = 0
for word in name:
word = word.lower()
if word in word_vecs:
ent_vec[i] += word_vecs[word]
k += 1
for idx in range(len(word) - 1):
char_vec[i, char2id[word[idx:idx + 2]]] += 1
if k:
ent_vec[i] /= k
else:
ent_vec[i] = np.random.random(300) - 0.5
if np.sum(char_vec[i]) == 0:
char_vec[i] = np.random.random(len(char2id)) - 0.5
ent_vec[i] = ent_vec[i] / np.linalg.norm(ent_vec[i])
char_vec[i] = char_vec[i] / np.linalg.norm(char_vec[i])
with open(save_path_name, 'wb') as f:
pickle.dump(ent_vec, f)
with open(save_path_char, 'wb') as f:
pickle.dump(char_vec, f)
print("save entity emb done. ")
return ent_vec, char_vec
def visual_pivot_induction(args, left_ents, right_ents, img_features, ills):
l_img_f = img_features[left_ents] # left images
r_img_f = img_features[right_ents] # right images
img_sim = l_img_f.mm(r_img_f.t())
topk = args.img_ill_k
two_d_indices = get_topk_indices(img_sim, topk * 100)
del l_img_f, r_img_f, img_sim
visual_links = []
used_inds = []
count = 0
for ind in two_d_indices:
if left_ents[ind[0]] in used_inds:
continue
if right_ents[ind[1]] in used_inds:
continue
used_inds.append(left_ents[ind[0]])
used_inds.append(right_ents[ind[1]])
visual_links.append((left_ents[ind[0]], right_ents[ind[1]]))
count += 1
if count == topk:
break
count = 0.0
for link in visual_links:
if link in ills:
count = count + 1
print(f"{(count / len(visual_links) * 100):.2f}% in true links")
print(f"visual links length: {(len(visual_links))}")
train_ill = np.array(visual_links, dtype=np.int32)
return train_ill
def read_raw_data(file_dir, lang=[1, 2]):
"""
Read DBP15k/DWY15k dataset.
Parameters
----------
file_dir: root of the dataset.
Returns
-------
ent2id_dict : A dict mapping from entity name to ids
ills: inter-lingual links (specified by ids)
triples: a list of tuples (ent_id_1, relation_id, ent_id_2)
r_hs: a dictionary containing mappings of relations to a list of entities that are head entities of the relation
r_ts: a dictionary containing mappings of relations to a list of entities that are tail entities of the relation
ids: all ids as a list
"""
print('loading raw data...')
def read_file(file_paths):
tups = []
for file_path in file_paths:
with open(file_path, "r", encoding="utf-8") as fr:
for line in fr:
params = line.strip("\n").split("\t")
tups.append(tuple([int(x) for x in params]))
return tups
def read_dict(file_paths):
ent2id_dict = {}
ids = []
for file_path in file_paths:
id = set()
with open(file_path, "r", encoding="utf-8") as fr:
for line in fr:
params = line.strip("\n").split("\t")
ent2id_dict[params[1]] = int(params[0])
id.add(int(params[0]))
ids.append(id)
return ent2id_dict, ids
ent2id_dict, ids = read_dict([file_dir + "/ent_ids_" + str(i) for i in lang])
ills = read_file([file_dir + "/ill_ent_ids"])
triples = read_file([file_dir + "/triples_" + str(i) for i in lang])
r_hs, r_ts = {}, {}
for (h, r, t) in triples:
if r not in r_hs:
r_hs[r] = set()
if r not in r_ts:
r_ts[r] = set()
r_hs[r].add(h)
r_ts[r].add(t)
assert len(r_hs) == len(r_ts)
return ent2id_dict, ills, triples, r_hs, r_ts, ids
def loadfile(fn, num=1):
print('loading a file...' + fn)
ret = []
with open(fn, encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
x = []
for i in range(num):
x.append(int(th[i]))
ret.append(tuple(x))
return ret
def get_ids(fn,file_dir):
ids = []
id2name = {}
fbid2name = {}
if 'FB' in fn:
with open(os.path.join(file_dir, 'fbid2name.txt'), encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
fbid2name[th[0]] = th[1]
with open(fn, encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
ids.append(int(th[0]))
name = th[1]
if '<http://yago-knowledge.org/resource/' in name:
name = name[1:-1].split('/')[-1]
if 'FB' in fn:
if name in fbid2name:
name = fbid2name[name]
if '<http://dbpedia.org/resource/' in name:
name = name[1:-1].split('/')[-1].replace('_', ' ')
id2name[int(th[0])] = name
return ids, id2name
def get_id2rel(fn):
id2rel = {}
with open(fn, encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
rel = th[1]
if '/' in rel:
rel = rel.split('/')[-1]
id2rel[int(th[0])] = rel
return id2rel
def get_ent2id(fns):
ent2id = {}
for fn in fns:
with open(fn, 'r', encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
ent2id[th[1]] = int(th[0])
return ent2id
def split_camel_case(input_string):
words = re.findall(r'[A-Z]?[a-z]+|[A-Z]+(?=[A-Z]|$)', input_string)
return words
def db_str(s):
return ' '.join(split_camel_case(s[1:-1].split('/')[-1].replace('_',' ')))
def db_time(s):
s = s.split("^^")[0][1:-1]
if 'e' in s:
return s
if '-' not in s[1:]:
return s
s = s.split('-')
y = int(s[0].replace('#','0'))
m = int(s[1]) if s[1]!='##'else 1
d = int(s[2]) if s[2]!='##' and s[2]!='' else 1
return y + (m-1)/12 +(d-1)/30/12
def dbp_str(s):
if '<'==s[0] and '>'==s[-1]:
s = s[1:-1]
t = s.split('/')[-1].replace('_',' ')
t_ = ' '.join(split_camel_case(t))
if t_ == '':
return t
return t_
def dbp_value(s):
# print(s)
if '^^' in s:
s = s.split("^^")[0]
if ('<' == s[0] and '>' == s[-1]) or ('\"' == s[0] and '\"' == s[-1]):
s = s[1:-1]
elif '@' in s and s.index('@')>0:
s = '@'.join(s.split('@')[:-1])
if ('<' == s[0] and '>' == s[-1]) or ('\"' == s[0] and '\"' == s[-1]):
s = s[1:-1]
# print(s)
if s[-1]=='\"':
s = s[:-1]
else:
if ('<' == s[0] and '>' == s[-1]) or ('\"' == s[0] and '\"' == s[-1]):
s = s[1:-1]
return s
if 'e' in s:
return s
if '-' not in s[1:]:
return s
try:
s_ = s.split('-')
y = int(s_[0].replace('#','0'))
m = int(s_[1]) if s_[1]!='##'else 1
d = int(s_[2]) if s_[2]!='##' and s_[2]!='' else 1
return y + (m-1)/12 +(d-1)/30/12
except:
return s
def load_attr_withNums(datas,fns, ent2id_dict, file_dir, topk=0):
ans = [load_attr_withNum(data,fn,ent2id_dict) for data,fn in zip(datas,fns)]
if topk!=0:
rels = []
rels2index = {}
rels2times = {}
cur = 0
att2rel = []
for i, att in enumerate(ans[0]+ans[1]):
if att[1] not in rels2index:
rels2index[att[1]] = cur
rels.append(att[1])
cur += 1
rels2times[att[1]] = 0
rels2times[att[1]] += 1
att2rel.append(rels2index[att[1]])
att2rel = np.array(att2rel)
rels_left = []
rels2index_left = {}
cur = 0
att2rel_left = []
for i, att in enumerate(ans[0]):
if att[1] not in rels2index_left:
rels2index_left[att[1]] = cur
rels_left.append(att[1])
cur += 1
att2rel_left.append(rels2index_left[att[1]])
att2rel_left = np.array(att2rel_left)
rels_right = []
rels2index_right = {}
cur = 0
att2rel_right = []
for i, att in enumerate(ans[1]):
if att[1] not in rels2index_right:
rels2index_right[att[1]] = cur
rels_right.append(att[1])
cur += 1
att2rel_right.append(rels2index_right[att[1]])
att2rel_right = np.array(att2rel_right)
rels_right = set(rels_right)
rels_left = set(rels_left)
rels_inter = rels_left.intersection(rels_right)
if len(rels_inter)==0:
rels_inter = rels
# select topk
rels_inter = sorted(rels_inter, key=lambda x: rels2times[x], reverse=True)[:topk]
ans_ = []
for i in ans[0]:
if i[1] in rels_inter:
ans_.append(i)
num_left = len(ans_)
for i in ans[1]:
if i[1] in rels_inter:
ans_.append(i)
num_right = len(ans_)-num_left
return ans_,num_left,num_right
# num_att_left = len(rels2index)
# att_rel_features = np.load(os.path.join(file_dir, 'att_rel_features.npy'), allow_pickle=True)
# rels = torch.FloatTensor(att_rel_features).cuda()
# sim_rels_left = torch.mm(rels[:num_att_left], rels[num_att_left:].T)
# sim_rels_right = torch.mm(rels[num_att_left:], rels[:num_att_left].T)
# # get the max sim at row
# sim_rels_left = torch.max(sim_rels_left, dim=1)[0]
# sim_rels_right = torch.max(sim_rels_right, dim=1)[0]
# # get the topk rels
# topk_rels_left = torch.topk(sim_rels_left, topk, dim=0)[1]
# topk_rels_right = torch.topk(sim_rels_right, topk, dim=0)[1]
#
# topk_rels_left = topk_rels_left.cpu().numpy()
# topk_rels_right = topk_rels_right.cpu().numpy()
# # topk_rels = np.concatenate([topk_rels_left,topk_rels_right+num_att_left])
#
#
# # contain topkrels
# common_elements = np.in1d(att2rel, topk_rels_left)
# common_elements_indices = list(np.where(common_elements)[0])
# ans_ = []
# for i in common_elements_indices:
# ans_.append(ans[0][i])
# num_left = len(ans_)
#
# rels = []
# rels2index = {}
# cur = 0
# att2rel = []
# for i,att in enumerate(ans[1]):
# if att[1] not in rels2index:
# rels2index[att[1]] = cur
# rels.append(att[1])
# cur += 1
# att2rel.append(rels2index[att[1]])
# att2rel = np.array(att2rel)
# # contain topkrels
# common_elements = np.in1d(att2rel, topk_rels_right)
# common_elements_indices = list(np.where(common_elements)[0])
# for i in common_elements_indices:
# ans_.append(ans[1][i])
# num_right = len(ans_) - num_left
# return ans_,num_left,num_right
return ans[0]+ans[1], len(ans[0]), len(ans[1])
def load_attr_withNum(data, fn, ent2id):
with open(fn, 'r',encoding='utf-8') as f:
Numericals = f.readlines()
if data == 'FB15K' or data == 'DB15K' or data=='YAGO15K':
Numericals_ = list(set(Numericals))
Numericals_.sort(key = Numericals.index)
Numericals = Numericals_
if data=='FB15K':
Numericals = [i[:-1].split('\t') for i in Numericals]
Numericals = [(ent2id[i[0]], i[1][1:-1].replace('http://rdf.freebase.com/ns/', '').split('.')[-1].replace('_',' '), i[2]) for i in
Numericals]
elif data=='DB15K':
Numericals = [i[:-1].split(' ') if '\t' not in i else i[:-1].split('\t') for i in Numericals]
Numericals = [(ent2id[i[0]], db_str(i[1]), db_time(i[2])) for i in Numericals]
elif data=='YAGO15K':
Numericals = [i[:-1].split(' ') if '\t' not in i else i[:-1].split('\t') for i in Numericals]
Numericals = [(ent2id[i[0]], db_str(i[1]), db_time(i[2])) for i in Numericals]
elif data=='oea':
Numericals = [i[:-1].split('\t') for i in Numericals]
Numericals = [(ent2id[i[0]], dbp_str(i[1]), dbp_value(i[2])) for i in Numericals]
else:
Numericals = [i[:-1].split(' ') if '\t' not in i else i[:-1].split('\t') for i in Numericals]
Numericals = [(ent2id[i[0][1:-1]], dbp_str(i[1]), dbp_value(' '.join(i[2:]))) for i in Numericals]
return Numericals
# The most frequent attributes are selected to save space
def load_attr(fns, e, ent2id, topA=1000):
cnt = {}
for fn in fns:
with open(fn, 'r', encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
if th[0] not in ent2id:
continue
for i in range(1, len(th)):
if th[i] not in cnt:
cnt[th[i]] = 1
else:
cnt[th[i]] += 1
fre = [(k, cnt[k]) for k in sorted(cnt, key=cnt.get, reverse=True)]
attr2id = {}
# pdb.set_trace()
topA = min(1000, len(fre))
for i in range(topA):
attr2id[fre[i][0]] = i
attr = np.zeros((e, topA), dtype=np.float32)
for fn in fns:
with open(fn, 'r', encoding='utf-8') as f:
for line in f:
th = line[:-1].split('\t')
if th[0] in ent2id:
for i in range(1, len(th)):
if th[i] in attr2id:
attr[ent2id[th[0]]][attr2id[th[i]]] = 1.0
return attr
def load_relation(e, KG, topR=1000):
# (39654, 1000)
rel_mat = np.zeros((e, topR), dtype=np.float32)
rels = np.array(KG)[:, 1]
top_rels = Counter(rels).most_common(topR)
rel_index_dict = {r: i for i, (r, cnt) in enumerate(top_rels)}
for tri in KG:
h = tri[0]
r = tri[1]
o = tri[2]
if r in rel_index_dict:
rel_mat[h][rel_index_dict[r]] += 1.
rel_mat[o][rel_index_dict[r]] += 1.
return np.array(rel_mat)
def load_json_embd(path):
embd_dict = {}
with open(path) as f:
for line in f:
example = json.loads(line.strip())
vec = np.array([float(e) for e in example['feature'].split()])
embd_dict[int(example['guid'])] = vec
return embd_dict
def load_img(e_num, path):
img_dict = pickle.load(open(path, "rb"))
# init unknown img vector with mean and std deviation of the known's
imgs_np = np.array(list(img_dict.values()))
mean = np.mean(imgs_np, axis=0)
std = np.std(imgs_np, axis=0)
# img_embd = np.array([np.zeros_like(img_dict[0]) for i in range(e_num)]) # no image
# img_embd = np.array([img_dict[i] if i in img_dict else np.zeros_like(img_dict[0]) for i in range(e_num)])
img_embd = np.array([img_dict[i] if i in img_dict else np.random.normal(mean, std, mean.shape[0]) for i in range(e_num)])
print(f"{(100 * len(img_dict) / e_num):.2f}% entities have images")
return img_embd