-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdata.py
1327 lines (1214 loc) · 62.1 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
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
from torch.utils.data import DataLoader, random_split
import numpy as np
from copy import deepcopy
import pickle
import pandas as pd
import torch
import os
import random
class Data:
def __init__(self, args=None):
neg_data_type = args.negative_triple_generation
data_dir = args.path_dataset_folder
emb_typ = args.emb_type
valid_ratio = args.valid_ratio
selected_dataset_data_dir = data_dir+str(args.eval_dataset).lower()+"/"
tmp_emb_folder = data_dir + str(args.eval_dataset).lower()+"/embeddings/"
ids_only = args.ids_only
# Quick workaround as we happen to have duplicate triples.
# None if load complete data, otherwise load parts of dataset with folders in wrong directory.
# emb_folder = ""
if args.model == "KGE-only":
self.process_KGE_only_data(selected_dataset_data_dir, args)
elif ids_only == False:
self.train_set_time_final = list((self.load_data(selected_dataset_data_dir + "train/", data_type="train")))
self.test_set_time_final = list((self.load_data(selected_dataset_data_dir + "test/", data_type="test")))
self.test_set_time_final, self.valid_set_time_final = self.generate_test_valid_set(self,
self.test_set_time_final, valid_ratio)
if args.include_veracity == True:
# factcheck predictions on train and test data. Should be added here before: 'data_TP/dbpedia124k/factcheck_veracity_scores/train_pred'
self.train_set_pred = list((self.load_data(selected_dataset_data_dir + "train/",
data_type="train_v_scores.txt", pred=True)))
self.test_set_pred = list((self.load_data(selected_dataset_data_dir + "test/",
data_type="test_v_scores.txt", pred=True)))
self.test_set_pred, self.valid_set_pred = self.generate_test_valid_set(self, self.test_set_pred, valid_ratio)
# generate test and validation sets
# self.test_set, self.valid_set = self.generate_test_valid_set(self, self.test_set)
###########################################################################################################
##########################################################################################################
##################SENTENCE WORLD###########################################################
self.emb_sentences_train = pd.read_csv(selected_dataset_data_dir + "train/" + "trainSE.csv", sep=",").iloc[:, 3:]
self.emb_sentences_test = pd.read_csv(selected_dataset_data_dir + "test/" + "testSE.csv", sep=",").iloc[:, 3:]
self.emb_sentences_test, self.emb_sentences_valid = self.generate_test_valid_sentence_set(self, self.emb_sentences_test, valid_ratio)
#############################################################################################################
##############################################################################################################
# get all entities and relations
# self.data = self.train_set + list(self.test_set) + list(self.valid_set)
self.data = self.train_set_time_final + self.test_set_time_final + self.valid_set_time_final
self.entities = self.get_entities(self.data)
# self.relations = list(set(self.get_relations(self.train_set) + self.get_relations(self.test_set)))
self.relations = self.get_relations(self.data)
self.times = self.get_times(self.data)
# self.save_all_resources(self.entities, selected_dataset_data_dir, is_entity=True)
# self.save_all_resources(self.relations, selected_dataset_data_dir, is_entity=False)
# exit(1)
self.idx_ent_dict = dict()
self.idx_rel_dict = dict()
self.idx_time_dict = dict()
# Generate integer mapping
for i in self.entities:
self.idx_ent_dict[i.replace("<http://dbpedia.org/resource/", "")[:-1]] = len(self.idx_ent_dict)
for i in self.relations:
self.idx_rel_dict[i.replace("<http://dbpedia.org/ontology/", "")[:-1]] = len(self.idx_rel_dict)
for i in self.times:
self.idx_time_dict[i] = len(self.idx_time_dict)
if args.include_veracity == True:
self.copaal_veracity_train = self.get_veracity_data(self, self.train_set_pred)
self.copaal_veracity_test = self.get_veracity_data(self, self.test_set_pred)
self.copaal_veracity_valid = self.get_veracity_data(self, self.valid_set_pred)
self.emb_entities = self.get_embeddings(tmp_emb_folder+emb_typ+'/','all_entities_embeddings_final.csv')
self.emb_relation = self.get_embeddings(tmp_emb_folder+emb_typ+'/','all_relations_embeddings_final.csv')
if str(args.model).__contains__("temporal"):
self.emb_time = self.get_embeddings(tmp_emb_folder+emb_typ+'/','time.pkl')
self.num_entities = len(self.emb_entities)
self.num_relations = len(self.emb_relation)
self.num_times = 0
if str(args.model).__contains__("temporal"):
self.num_times = len(self.emb_time)
if args.negative_triple_generation =="corrupted-time-based":
self.train_set_time_final = self.generate_negative_triples(self.train_set_time_final)
self.valid_set_time_final = self.generate_negative_triples(self.valid_set_time_final)
self.test_set_time_final = self.generate_negative_triples(self.test_set_time_final)
elif args.negative_triple_generation == "False":
self.train_set_time_final = self.generate_only_true_triples(self.train_set_time_final)
self.valid_set_time_final = self.generate_only_true_triples(self.valid_set_time_final)
self.test_set_time_final = self.generate_only_true_triples(self.test_set_time_final)
self.idx_train_set = []
i = 0
sent_i = 0
len_train = len(self.train_set_time_final)
for (s, p, o, time, label) in self.train_set_time_final:
s = str(s).replace("<http://dbpedia.org/resource/", "")[:-1]
p = str(p).replace("<http://dbpedia.org/ontology/", "")[:-1].replace("Of","")
o = str(o).replace("<http://dbpedia.org/resource/", "")[:-1]
if self.idx_ent_dict.keys().__contains__(s) and self.idx_rel_dict.keys().__contains__(p) and self.idx_ent_dict.keys().__contains__(o):
idx_s, idx_p, idx_o, idx_t, label = self.idx_ent_dict[s], self.idx_rel_dict[p], self.idx_ent_dict[o], self.idx_time_dict[time], label
if label == 'True' or label == 1:
label = 1
else:
label = 0
ver = i
# this is to just to make sure if any time is not the same even after randomaly shuffling so increment by 1
if args.negative_triple_generation =="corrupted-time-based" and (i >= int(len_train/2)):
item = self.idx_train_set.__getitem__(i-int(len_train/2))
if ((item[0] != int(idx_s)) or (item[1] != int(idx_p)) or (item[2]!=int(idx_o))):
print("serious problem, please check")
exit(1)
if ((item[0]==int(idx_s)) and (item[1]==int(idx_p)) and (item[2]==int(idx_o)) and (item[3]==int(idx_t))):
idx_t = (int(idx_t) + 1) if ((int(idx_t)+1) < self.num_times) else 0
ver = item[4]
self.idx_train_set.append([int(idx_s), int(idx_p), int(idx_o),int(idx_t), sent_i, ver, label])
else:
print("check:"+s + ","+o)
i = i + 1
sent_i = sent_i + 1
self.idx_valid_set = []
j = 0
sent_i = 0
len_valid = len(self.valid_set_time_final)
for (s, p, o, time, label) in self.valid_set_time_final:
s = str(s).replace("<http://dbpedia.org/resource/", "")[:-1]
p = str(p).replace("<http://dbpedia.org/ontology/", "")[:-1].replace("Of","")
o = str(o).replace("<http://dbpedia.org/resource/", "")[:-1]
if self.idx_ent_dict.keys().__contains__(s) and self.idx_rel_dict.keys().__contains__(p) and self.idx_ent_dict.keys().__contains__(o):
idx_s, idx_p, idx_o, idx_t, label = self.idx_ent_dict[s], self.idx_rel_dict[p], self.idx_ent_dict[o],self.idx_time_dict[time], label
if label == 'True' or label == 1:
label = 1
else:
label = 0
ver = j
# this is to check if any time is same even after randomaly shuffling so increment by 1
if args.negative_triple_generation =="corrupted-time-based" and (j >= int(len_valid/2)):
item = self.idx_valid_set.__getitem__(j-int(len_valid/2))
if ((item[0] != int(idx_s)) or (item[1] != int(idx_p)) or (item[2]!=int(idx_o))):
print("serious problem, please check")
exit(1)
if ((item[0]==int(idx_s)) and (item[1]==int(idx_p)) and (item[2]==int(idx_o)) and (item[3]==int(idx_t))):
idx_t = (int(idx_t) + 1) if ((int(idx_t)+1) < self.num_times) else 0
ver = item[4]
self.idx_valid_set.append([int(idx_s), int(idx_p), int(idx_o),int(idx_t), sent_i, ver, label])
else:
print("check:" + s + "," + o)
j = j + 1
sent_i = sent_i + 1
self.idx_test_set = []
k = 0
sent_i = 0
len_test = len(self.test_set_time_final)
for (s, p, o, time, label) in self.test_set_time_final:
s = str(s).replace("<http://dbpedia.org/resource/", "")[:-1]
p = str(p).replace("<http://dbpedia.org/ontology/", "")[:-1].replace("Of","")
o = str(o).replace("<http://dbpedia.org/resource/", "")[:-1]
if self.idx_ent_dict.keys().__contains__(s) and self.idx_rel_dict.keys().__contains__(p) and self.idx_ent_dict.keys().__contains__(o):
idx_s, idx_p, idx_o, idx_t, label = self.idx_ent_dict[s], self.idx_rel_dict[p], self.idx_ent_dict[o],self.idx_time_dict[time], label
if label == 'True' or label == 1:
label = 1
else:
label = 0
ver = k
# this is to check if any time is same even after randomaly shuffling so increment by 1
if args.negative_triple_generation =="corrupted-time-based" and (k >= int(len_test/2)):
item = self.idx_test_set.__getitem__(k-int(len_test/2))
if ((item[0] != int(idx_s)) or (item[1] != int(idx_p)) or (item[2]!=int(idx_o))):
print("serious problem, please check")
exit(1)
if ((item[0]==int(idx_s)) and (item[1]==int(idx_p)) and (item[2]==int(idx_o)) and (item[3]==int(idx_t))):
idx_t = (int(idx_t) + 1) if ((int(idx_t)+1) < self.num_times) else 0
ver = item[4]
self.idx_test_set.append([int(idx_s), int(idx_p), int(idx_o),int(idx_t), sent_i, ver, label])
else:
print("check:" + s + "," + o)
k = k + 1
sent_i = sent_i + 1
else:
# self.idx_ent_dict = self.get_ids_dict(selected_dataset_data_dir+"entities")
# self.idx_rel_dict = self.get_ids_dict(selected_dataset_data_dir+"relations")
# self.idx_time_dict = self.get_ids_dict(selected_dataset_data_dir+"times")
self.emb_entities = self.get_embeddings( tmp_emb_folder + emb_typ + '/', 'entity.pkl')
self.emb_relation = self.get_embeddings( tmp_emb_folder + emb_typ + '/', 'relation.pkl')
self.emb_time = self.get_embeddings( tmp_emb_folder + emb_typ + '/', 'time.pkl')
self.num_entities = len(self.emb_entities)
self.num_relations = len(self.emb_relation)
self.num_times = len(self.emb_time)
self.idx_train_set = self.get_ids_dict(selected_dataset_data_dir+"train/train")
self.idx_test_set = self.get_ids_dict(selected_dataset_data_dir+"test/test")
self.idx_valid_set = self.get_ids_dict(selected_dataset_data_dir+"test/valid")
# Function to find a key by its value in a dictionary
def process_KGE_only_data(self, selected_dataset_data_dir, args, valid_ratio):
self.idx_train_set = []
self.idx_test_set = []
self.idx_valid_set = []
# reading train and test sets
self.train_set = list(
(self.load_data(selected_dataset_data_dir + "train/", data_type="train")))
self.test_set = list(
(self.load_data(selected_dataset_data_dir + "test/", data_type="test")))
self.test_set, self.valid_set = self.generate_test_valid_set(self, self.test_set, valid_ratio)
# negative triples generation
if args.negative_triple_generation != "corrupted-time-based":
self.train_set = self.generate_negative_triples(self.train_set, "corrupted-triple-based")
self.test_set = self.generate_negative_triples(self.test_set, "corrupted-triple-based")
self.valid_set = self.generate_negative_triples(self.valid_set, "corrupted-triple-based")
self.idx_ent_dict = dict()
self.idx_rel_dict = dict()
self.data = self.train_set + self.test_set + self.valid_set
self.entities = self.get_entities(self.data)
self.relations = self.get_relations(self.data)
# self.save_all_resources(self.entities, selected_dataset_data_dir, is_entity=True)
# self.save_all_resources(self.relations, selected_dataset_data_dir, is_entity=False)
# exit(1)
# Generate integer mapping
for i in self.entities:
self.idx_ent_dict[i] = len(self.idx_ent_dict)
for i in self.relations:
self.idx_rel_dict[i] = len(self.idx_rel_dict)
self.emb_entities = self.get_embeddings_from_csv(selected_dataset_data_dir + 'embeddings/'+args.emb_type,
'/entities_embeddings.csv', self.entities)
self.emb_relation = self.get_embeddings_from_csv(selected_dataset_data_dir + 'embeddings/'+args.emb_type,
'/relations_embeddings.csv', self.relations)
self.num_entities = len(self.emb_entities)
self.num_relations = len(self.emb_relation)
self.num_times = 0
# creating ids of the train andtest sets
i = 0
for (s, p, o, label) in self.train_set:
if self.idx_ent_dict.keys().__contains__(s) and self.idx_rel_dict.keys().__contains__(
p) and self.idx_ent_dict.keys().__contains__(o):
idx_s, idx_p, idx_o, label = self.idx_ent_dict[s], self.idx_rel_dict[p], self.idx_ent_dict[o], label
if label == 'True' or label == 1:
label = 1
else:
label = 0
ver = i
self.idx_train_set.append([int(idx_s), int(idx_p), int(idx_o), 0, ver, label])
else:
print("check:" + s + "," + o)
i = i + 1
i = 0
for (s, p, o, label) in self.test_set:
if self.idx_ent_dict.keys().__contains__(s) and self.idx_rel_dict.keys().__contains__(
p) and self.idx_ent_dict.keys().__contains__(o):
idx_s, idx_p, idx_o, label = self.idx_ent_dict[s], self.idx_rel_dict[p], self.idx_ent_dict[o], label
if label == 'True' or label == 1:
label = 1
else:
label = 0
ver = i
self.idx_test_set.append([int(idx_s), int(idx_p), int(idx_o), 0, ver, label])
else:
print("check:" + s + "," + o)
i = i + 1
i = 0
for (s, p, o, label) in self.valid_set:
if self.idx_ent_dict.keys().__contains__(s) and self.idx_rel_dict.keys().__contains__(
p) and self.idx_ent_dict.keys().__contains__(o):
idx_s, idx_p, idx_o, label = self.idx_ent_dict[s], self.idx_rel_dict[p], self.idx_ent_dict[o], label
if label == 'True' or label == 1:
label = 1
else:
label = 0
ver = i
self.idx_valid_set.append([int(idx_s), int(idx_p), int(idx_o), 0, ver, label])
else:
print("check:" + s + "," + o)
i = i + 1
print("loading train and test is done")
def get_key(self,dictionary, value):
for key, val in dictionary.items():
if val == value:
return key
return None
def generate_negative_triples(self, data, type="time-based"):
data2 = []
data_final = []
i =0
if type=="corrupted-time-based":
times = []
for (s, p, o, time, label) in data:
if label == 'True':
times.append(time)
data2.append([s, p, o, time, label])
i = i + 1
random.shuffle(times)
data3 = []
for j in range(len(data2)):
item = data2.__getitem__(j)
tim = times.__getitem__(j)
data3.append([item[0],item[1],item[2],tim,'False'])
data_final = data2 + data3
else:
relations = []
i = 0
for (s, p, o, label) in data:
if label == 'True' or label == 1:
relations.append(p)
data2.append([s, p, o, label])
i = i + 1
relations = set(relations)
idx_relations = dict()
for rel in relations:
idx_relations[rel] = len(idx_relations)
data3 = []
for j in range(len(data2)):
item = data2.__getitem__(j)
rr = idx_relations[item[1]]
new_idx = 0
if rr < len(idx_relations)-1:
new_idx = rr+1
new_r = self.get_key(idx_relations, new_idx)
data3.append([item[2], new_r, item[0], 0])
data_final = data2 + data3
# data_final.append(data3)
return data_final
def generate_only_true_triples(self, data):
data2 = []
data_final = []
i =0
times = []
for (s, p, o, time, label) in data:
if label == 'True\n':
label = label[:-1]
if label == 'True':
times.append(time)
data_final.append([s, p, o, time, label])
if label == 1:
times.append(time)
data_final.append([s, p, o, time, label])
i = i + 1
# data_final.append(data3)
return data_final
@staticmethod
def get_veracity_data(self, train_emb):
embeddings_train = dict()
i = 0
for train in train_emb:
embeddings_train[i] = float(str(train[3]).replace(".\n",""))
i += 1
return pd.DataFrame(embeddings_train.values())
@staticmethod
def update_and_match_triples_start(self, selected_dataset_data_dir, type, file_name, data_set1, data_set2, properties_split = None, veracity = False):
if veracity==False:
if (os.path.exists(selected_dataset_data_dir + type+ "/"+ file_name)):
self.set_time_final = list(self.load_data(selected_dataset_data_dir + type+"/", data_type=str(file_name).replace(".txt",""),pred=True))
else:
if len(data_set1) != len(data_set2):
self.set_time_final = self.update_match_triples(data_set1, data_set2)
else:
self.set_time_final = data_set2
self.save_triples(selected_dataset_data_dir, type+"/"+file_name, self.set_time_final)
else:
tt = "properties/train/" if (file_name.__contains__("train")) else "properties/test/"
split = "" if (properties_split==None) else tt+"correct/" +properties_split + "_"
if (os.path.exists(selected_dataset_data_dir + type+ "/"+ split+ file_name)):
self.set_time_final = list(self.load_data(selected_dataset_data_dir + type+"/"+split , data_type=str(file_name).replace(".txt",""),pred=True))
else:
self.set_time_final = self.update_match_triples(data_set1, data_set2, veracity=veracity)
self.save_triples(selected_dataset_data_dir, type+"/"+split+file_name, self.set_time_final,veracity=veracity)
return self.set_time_final
def is_valid_test_available(self):
if len(self.idx_valid_set) > 0 and len(self.idx_test_set) > 0:
return True
return False
# @staticmethod
# def load_triples(data_dir, type, triples):
# with open(data_dir + type + '.txt', "r") as f:
# for item in triples:
# f.write("%s\n" % item)
@staticmethod
def save_triples(data_dir,type, triples,veracity=False):
if veracity==False:
with open(data_dir + type, "w") as f:
for item in triples:
f.write(""+(item[0])+"\t"+(item[1])+"\t"+(item[2])+"\t"+str(item[3])+"\t"+str(item[4])+"\n")
else:
with open(data_dir + type, "w") as f:
for item in triples:
f.write(""+str(item[0])+"\t"+str(item[1])+"\t"+str(item[2])+"\t"+str(item[3])+"\n")
@staticmethod
def save_all_resources(list_all_entities, data_dir, sub_path="", is_entity=True):
if is_entity:
with open(data_dir+sub_path+'all_entities.txt',"w") as f:
for item in list_all_entities:
f.write("%s\n" % item)
else:
with open(data_dir + sub_path + 'all_relations.txt', "w") as f:
for item in list_all_entities:
f.write("%s\n" % item)
@staticmethod
def generate_test_valid_set(self, test_set, valid_ratio):
test_data = []
valid_data = []
i = 0
sent_i = 0
for data in test_set:
if i % valid_ratio == 0:
valid_data.append(data)
else:
test_data.append(data)
i += 1
return test_data, valid_data
@staticmethod
def generate_test_valid_sentence_set(self, test_set, valid_ratio):
valid_indices = list(range(0, len(test_set), valid_ratio)) # Indices for validation set
all_indices = list(range(len(test_set)))
test_indices = [idx for idx in all_indices if idx not in valid_indices] # Indices for test set
test_data = test_set.iloc[test_indices] # Extract test set
valid_data = test_set.iloc[valid_indices] # Extract validation set
return test_data, valid_data
def get_ids_dict(self, dict_file_path):
ids_dict = dict()
data = []
with open("%s" % (dict_file_path), "r") as f:
for datapoint in f:
datapoint = datapoint.split()
if len(datapoint) == 2:
ids_dict[datapoint[1]] = datapoint[0]
elif len(datapoint)==5:
arr = []
for tt in datapoint:
arr.append(int(tt))
arr.append(True)
data.append(arr)
else:
print("invalid format")
exit(1)
if len(ids_dict) > 0:
return ids_dict
else:
return data
@staticmethod
def load_data(data_dir, data_type, pred=False):
try:
data = []
if pred == False:
with open("%s%s" % (data_dir, data_type), "r") as f:
for datapoint in f:
datapoint = datapoint.split("\t")
if len(datapoint) == 4:
s, p, o, label = datapoint
if label == '.\n': # TODO if false triples are also provided then label could be false or??
label = 1
elif label == 'True' or label == 1 or label == '1':
label = 1
else:
label = 0
data.append((s, p, o, label))
elif len(datapoint) == 3:
s, p, label = datapoint
assert label == 'True' or label == 'False'
if label == 'True':
label = 1
else:
label = 0
data.append((s, p, 'DUMMY', label))
elif len(datapoint) == 5:
if datapoint[4]==".\n":
s, p, o, label, dummy = datapoint
label = label.replace("\n", "")
if label == 'True' or label == '1.0' or label == 1.0 or label == '1' or label == 1:
label = 1
else:
label = 0
data.append((s, p, o,"N/A", label))
else:
s, p, o, time, label = datapoint
label=label.replace("\n","")
assert label == 'True' or label == 'False'
if label == 'True' or label == '1' or label == 1:
label = 1
else:
label = 0
data.append((s, p, o, time, label))
else:
raise ValueError
else:
with open("%s%s" % (data_dir, data_type), "r") as f:
for datapoint in f:
datapoint = datapoint.split('\t')
if len(datapoint) == 4:
s, p, o, label = datapoint
label = str(label).replace("\n","")
data.append((s, p, o, float(label)))
elif len(datapoint) == 3:
s, p, label = datapoint
label = str(label).replace("\n", "")
data.append((s, p, 'DUMMY', float(label)))
elif len(datapoint) == 5:
s, p, o, time, label = datapoint
label = str(label).replace("\n", "")
if label=="." and str(time).__contains__("^<http://www.w3.org/2001/XMLSchema#double>"):
label = time
label = str(label).replace("\"^^<http://www.w3.org/2001/XMLSchema#double>", "")
data.append((s, p, o, float(label)))
elif label=="." and (str(time).startswith('0.') or time == '1.0'):
data.append((s, p, o, float(time)))
else:
data.append((s, p, o, time, float(label)))
else:
raise ValueError
except FileNotFoundError as e:
print(e)
print('Add empty.')
data = []
return data
@staticmethod
def get_mapped_entities(data_dir, file_name):
mapping_entities = dict()
with open("%s%s.txt" % (data_dir, file_name), "r") as f:
for datapoint in f:
datapoint = datapoint.split(" -> ")
if len(datapoint) == 2:
mapping_entities[datapoint[0]] = datapoint[1].replace("\n","")
return mapping_entities
def check_if_not_equal_size(self, data_set1, data_set2):
data = []
if len(data_set1)!=len(data_set2):
train_set_time1 = deepcopy(data_set1)
for tp in data_set2:
found = False
for tpt in train_set_time1:
if (tp[0] == tpt[0] and tp[1] == tpt[1].replace("Of", "") and tp[2] == tpt[2]):
data.append([tpt[0], tpt[1], tpt[2], tpt[3]])
found = True
break
elif (tp[2] == tpt[0] and tp[0] == tpt[2]):
data.append([tpt[0], tpt[1], tpt[2], tpt[3]])
found = True
break
if found == False:
print("Embeddings not found: excluded triple:" + str(tp))
else:
train_set_time1.remove(tpt)
return data
else:
return data_set1
# update date set 1 and match with dataset 2
@staticmethod
def update_match_triples(data_set1, data_set2, veracity=False, final= False):
data = []
data_set21 = deepcopy(data_set2)
# subs = [tp2[0] for tp2 in data_set2]
# preds = [tp2[1].replace("Of","") for tp2 in data_set2]
# objs = [tp2[2] for tp2 in data_set2]
for tp in data_set1:
found = False
if veracity == False:
for tpt in data_set21:
# if tpt[0].__contains__('Amadou_Toumani') and (tp[2].__contains__('Amadou_Toumani_')):
# print("test")
if (tp[0] == tpt[0] and tp[1] == tpt[1].replace("Of","") and tp[2] == tpt[2]):
data.append([tp[0],tp[1],tp[2],tpt[3],tpt[4]])
found = True
break
elif(tp[2] == tpt[0] and tp[0] == tpt[2]):# to cover negative triples we are doing like this, swaping the sub and obj and not checking the predicate
data.append([tpt[0], tpt[1], tpt[2], tpt[3], 'False'])
found = True
break
if found == False:
print("not found:"+ str(tp))
elif veracity==True and final==True: # final is for second check
for tpt in data_set21:
if (tp[0] == tpt[0] and tp[1] == tpt[1].replace("Of", "") and tp[2] == tpt[2]):
data.append([tp[0], tp[1], tp[2], tp[3], tp[4]])
found = True
break
elif (tp[2] == tpt[0] and tp[1] == tpt[1].replace("Of", "") and tp[0] == tpt[2]):# to cover negative triples we are doing like this, swaping the sub and obj and not checking the predicate
data.append([tp[0], tp[1], tp[2], tp[3], tp[4]])
found = True
break
if found == False:
print("not found:"+ str(tp))
else:
data_set21.remove(tpt)
else:
for tpt in data_set21:
if (tp[0] == tpt[0] and tp[1] == tpt[1].replace("Of", "") and tp[2] == tpt[2]):
data.append([tp[0], tp[1], tp[2], tp[3]])
found = True
break
elif (tp[2] == tpt[0] and tp[0] == tpt[2]): # to cover negative triples we are doing like this, swaping the sub and obj and not checking the predicate
data.append([tp[0], tp[1], tp[2], tp[3]])
found = True
break
if found == False:
print("not found:"+ str(tp))
# break
# else:
# print("problematic triple:"+ str(tp))
return data
@staticmethod
def load_data_with_time(data_dir, data_type, mapped_entities=None, prop = None):
try:
data = []
with open("%s%s.txt" % (data_dir, data_type), "r") as f:
for datapoint in f:
datapoint = datapoint.split("\t")
if len(datapoint) >= 5:
if len(datapoint) >5:
datapoint[5] = '_'.join(datapoint[4:])
s, p, o, time, loc = datapoint[0:5]
if prop!=None:
if not str(p).__eq__(prop+"Of"):
continue
s = "http://dbpedia.org/resource/" + s
if (mapped_entities!=None and s in mapped_entities.keys()):
s = mapped_entities[s]
p = "http://dbpedia.org/ontology/" + p
o = "http://dbpedia.org/resource/" + o
if (mapped_entities!=None and o in mapped_entities.keys()):
o = mapped_entities[o]
data.append(("<" + s + ">", "<" + p + ">", "<" + o + ">", time, "True"))
elif len(datapoint) == 3:
s, p, label = datapoint
assert label == 'True' or label == 'False'
if label == 'True':
label = 1
else:
label = 0
data.append((s, p, 'DUMMY', label))
else:
raise ValueError
except FileNotFoundError as e:
print(e)
print('Add empty.')
data = []
return data
@staticmethod
def get_relations(data):
relations = sorted(list(set([d[1] for d in data])))
return relations
@staticmethod
def get_entities(data):
entities = sorted(list(set([d[0] for d in data] + [d[2] for d in data])))
return entities
@staticmethod
def get_times(data):
times = sorted(list(set([d[3] for d in data])))
return times
# / home / umair / Documents / pythonProjects / HybridFactChecking / Embeddings / ConEx_dbpedia
@staticmethod
def get_embeddings_from_csv(path,name, order):
embd = pd.read_csv("%s%s" % (path, name))
embd['Key'] = pd.Categorical(embd['Key'], categories=order, ordered=True)
# Sort the DataFrame based on the 'Fruits' column
sorted_df = embd.sort_values(by='Key')
return sorted_df.iloc[:, 1:]
@staticmethod
def get_embeddings(path,name):
# embeddings = dict()
# print("%s%s.txt" % (path,name))
if name.endswith(".pkl"):
embd = torch.load("%s%s" % (path,name),map_location=torch.device('cpu'))
# old_data = pickle.load(file)
# with open("%s%s" % (path,name), 'rb') as f:
# data = pickle.load(f)
return embd.weight
elif name.endswith(".csv"):
embd = pd.read_csv("%s%s" % (path,name), sep=",")
last_column_name = embd.columns[-1]
if str(embd[last_column_name]).__contains__("]"):
embd[last_column_name] = embd[last_column_name].str.replace(']', '', regex=False)
return embd.iloc[:, 1:]
else:
print("invalid embeddings format. Please use .csv or .pkl fomat")
raise ValueError
# for emb in idxs:
# if emb not in embeddings.keys():
# print("this is missing in embeddings file:"+ emb)
# exit(1)
#
# if len(idxs) > len(embeddings):
# print("embeddings missing")
# exit(1)
# embeddings_final = dict()
# for emb in idxs.keys():
# if emb in embeddings.keys():
# embeddings_final[emb] = embeddings[emb]
# else:
# print('no embedding', emb)
# exit(1)
return embd.weight
@staticmethod
def get_comma_seperated_embeddings(idxs, path, name):
embeddings = dict()
# print("%s%s.txt" % (path,name))
with open("%s%s.txt" % (path, name), "r") as f:
for datapoint in f:
data = datapoint.split('> ,')
if datapoint.startswith("<http://dbpedia.org/resource/Abu_Jihad_("):
print(datapoint)
if len(data) == 1:
data = datapoint.split('>\",')
if len(data) > 1:
data2 = data[0] + ">", data[1].split(',')
# test = data2[0].replace("\"","").replace("_com",".com").replace("Will-i-am","Will.i.am").replace("Will_i_am","Will.i.am")
test = data2[0].replace("\"", "")
if test in idxs:
embeddings[test] = data2[1]
# else:
# print('Not in embeddings:',datapoint)
# exit(1)
# else:
# print('Not in embeddings:',datapoint)
# exit(1)
for emb in idxs:
if emb not in embeddings.keys():
print("this is missing in embeddings file:" + emb)
exit(1)
if len(idxs) > len(embeddings):
print("embeddings missing")
exit(1)
embeddings_final = dict()
for emb in idxs.keys():
if emb in embeddings.keys():
embeddings_final[emb] = embeddings[emb]
else:
print('no embedding', emb)
exit(1)
return embeddings_final.values()
@staticmethod
def get_copaal_veracity(path, name, train_set):
emb = dict()
embeddings_train = dict()
# print("%s%s" % (path,name))
i = 0
train_i = 0
found = False
with open("%s%s" % (path, name), "r") as f:
for datapoint in f:
if datapoint.startswith("0,1,2"):
continue
else:
emb[i] = datapoint.split(',')
try:
for dd in train_set:
# figure out some way to handle this first argument well
if (emb[i][0] == dd[0].replace(',', '')) and (emb[i][1] == dd[1].replace(',', '')) and (
emb[i][2] == dd[2].replace(',', '')):
# print('train data found')
embeddings_train[train_i] =np.append(emb[i][:3],emb[i][-1].replace("\n",""))
train_i += 1
found = True
break
# else:
# print('error')
# exit(1)
except:
print('ecception')
exit(1)
if found == False:
if (train_i >= len(train_set)):
break
else:
print("some training data missing....not found:" + str(emb[i]))
exit(1)
i = i + 1
found = False
# i = i+1
embeddings_train_final = dict()
jj = 0
# print("sorting")
for embb in train_set:
ff = False
for embb2 in embeddings_train.values():
if ((embb[0].replace(',', '') == embb2[0].replace(',', '')) and (
embb[1].replace(',', '') == embb2[1].replace(',', '')) and (
embb[2].replace(',', '') == embb2[2].replace(',', ''))):
embeddings_train_final[jj] = embb2
jj = jj + 1
ff = True
break
if ff == False:
print("problem: not found")
exit(1)
if len(train_set) != len(embeddings_train_final):
print("problem")
exit(1)
return embeddings_train_final.values()
@staticmethod
def update_entity(self, ent):
ent = ent.replace("+", "")
if (ent.__contains__("&") or ent.__contains__("%")) and (
(not ent.__contains__("%3F")) and (not ent.__contains__("%22"))):
sub2 = ""
for chr in ent:
if chr == "&" or chr == "%":
break
else:
sub2 += chr
if ent[0]=="<":
ent = sub2 + ">"
else:
ent = sub2
if ent.__contains__("?"):
ent = ent.replace("?", "%3F")
if ent.__contains__("\"\""):
ent= ent.replace("\"\"", "%22")
if ent[0] == "\"" and ent[-1] == "\"":
ent = ent[1:-1]
if ent[0] == "\'" and ent[-1] == "\'":
ent = ent[1:-1]
return ent
def without(self,d, key):
new_d = d.copy()
new_d2 = dict()
new_d.pop(key)
count = 0
for dd in new_d.values():
new_d2[count]=dd
count+=1
return new_d2
@staticmethod
def get_sent_embeddings(self, path, name, train_set):
emb = dict()
embeddings_train = dict()
# print("%s%s" % (path,name))
train_set_copy = deepcopy(train_set)
i = 0
train_i = 0
found = False
with open("%s%s" % (path, name), "r") as f:
for datapoint in f:
if datapoint.startswith("0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20"):
continue
else:
if datapoint.startswith("http://dbpedia.org/resource/Vlado_Brankovic"):
print("test")
emb[i] = datapoint.split('\t')
try:
if emb[i][0] != "0":
for dd in train_set_copy:
# updated because factcheck results does not contained punctuations
sub = self.update_entity(self, dd[0])
pred = self.update_entity(self, dd[1])
obj = self.update_entity(self, dd[2])
emb[i][0] = self.update_entity(self, emb[i][0])
emb[i][1] = self.update_entity(self, emb[i][1])
emb[i][2] = self.update_entity(self, emb[i][2])
if ((emb[i][0] == sub) and
(emb[i][1] == pred) and
(emb[i][2] == obj)
or
('<'+emb[i][0].lower()+'>' == sub.lower()) and
('<'+emb[i][1].lower()+'>' == pred.lower()) and
('<'+emb[i][2].lower()+'>' == obj.lower())):
# print('train data found')
emb[i][-1] = emb[i][-1].replace("'", "").replace("\n","")
if (len(emb[i])) == ((768 * 3) +3 + 1):
# because defacto scores are also appended at the end
embeddings_train[train_i] = emb[i][:-1]
elif (len(emb[i])) == ((768 * 3) +3):
# emb[i][-1] = emb[i]
embeddings_train[train_i] = emb[i]
else:
print("there is something fishy:"+str(emb[i]))
exit(1)
train_i += 1
found = True
break
# else:
# print('error')
# exit(1)
except:
print('ecception')
exit(1)
if found==True:
train_set_copy.remove(dd)
# else:
# train_set.remove(dd)
if found == False:
if (train_i >= len(train_set)):
break
else:
print("some training data missing....not found:" + str(emb[i]))
print(i)
print("test")
# exit(1)
i = i + 1
found = False
# i = i+1
if len(train_set) != len(embeddings_train):
print("problem: length of train and sentence embeddings arrays are different:train:"+str(len(train_set))+",emb:"+str(len(embeddings_train)))
# exit(1)
# following code is just for ordering the data in sentence vectors
train_i = 0
train_set_copy = deepcopy(train_set)
embeddings_train_final = dict()
for dd in train_set:
found_data = False
jj = 0
for sd in embeddings_train.values():
sub = self.update_entity(self, dd[0])
pred =self.update_entity(self, dd[1])
obj = self.update_entity(self, dd[2])
sub1 = '<'+self.update_entity(self, sd[0])+'>'
pred1 = '<'+self.update_entity(self, sd[1])+'>'
obj1 = '<'+self.update_entity(self, sd[2])+'>'
if (((sub == sub1) and (pred == pred1) and (obj == obj1))
or
(( sub1.lower()== sub.lower()) and ( pred1.lower() == pred.lower()) and
( obj1.lower() == obj.lower()))):
embeddings_train_final[train_i] = sd
train_i+=1
found_data = True
break
jj += 1
if found_data== False:
train_set_copy.remove(dd)
print("missing train data from sentence embeddings file:"+str(dd))
else:
# print("to delete from list: "+str(sd))
embeddings_train = self.without(embeddings_train,jj)
# embeddings_train = embeddings_train.dropna().reset_index(drop=True)
# del embeddings_train[sd]
train_set = deepcopy(train_set_copy)
return embeddings_train_final.values(), train_set