-
Notifications
You must be signed in to change notification settings - Fork 1
/
main.py
1221 lines (1014 loc) · 48.4 KB
/
main.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
import math
import numpy as np
import torch
import os
from causal import SessionDataset, collate_fn, Recommender
import pickle
from pathlib import Path
import time
import argparse
from torch.utils.data import DataLoader
from tqdm import tqdm
# from pyinstrument import Profiler
def main():
# # todo
# profiler = Profiler()
# profiler.start()
#! 处理命令行参数
parser = argparse.ArgumentParser()
parser.add_argument("--d", default=' ', help='comments')
parser.add_argument("--data", default='delicious')
parser.add_argument("--split", default=0, type=int)
parser.add_argument("--load", default=False)
parser.add_argument("--save_path", default='model.pkl')
parser.add_argument("--sampling", default='recent')
parser.add_argument("--sample_size", default=10, type=int)
parser.add_argument("--max_len_recent", default=10, type=int)
parser.add_argument("--c", default=1, type=float)
parser.add_argument("--b1", default=0.1, type=float)
parser.add_argument("--b2", default=1, type=float)
parser.add_argument("--dim", default=64, type=int)
parser.add_argument("--n_epoch", default=20, type=int)
parser.add_argument("--patience", default=5, type=int)
parser.add_argument("--batch_size", default=64, type=int)
parser.add_argument("--lr", default=1e-2, type=float)
parser.add_argument("--l2", default=1e-5, type=float)
parser.add_argument("--eval_every", default=5,
help='evalutate every 5 epochs', type=int)
parser.add_argument("--no_cuda", default=False)
parser.add_argument("--device", default='cuda:0')
parser.add_argument("--seed", default=2022)
parser.add_argument("--weighting", default='div')
parser.add_argument("--user_key", default='user')
parser.add_argument("--item_key", default='item')
parser.add_argument("--time_key", default='timestamp')
parser.add_argument("--session_key", default='sessionId')
args = parser.parse_args()
#! 初始化其他参数
np.random.seed(args.seed)
torch.manual_seed(args.seed)
torch.cuda.manual_seed(args.seed)
torch.backends.cudnn.enabled = True
torch.backends.cudnn.benchmark = True
device = torch.device(args.device if torch.cuda.is_available()
and not args.no_cuda else 'cpu')
PRJ_PATH = Path(__file__).resolve().parent
DATA_PATH = PRJ_PATH / 'data'
#! dataset处理
# stime = time.time()
with open(DATA_PATH/f'fm_{args.data}_{args.split}.pkl', 'rb') as f:
trainset, testset = pickle.load(f)
# todo test code
load_dataset = False
if not load_dataset:
dataset = SessionDataset(trainset, testset, user_key=args.user_key, item_key=args.item_key, session_key=args.session_key,
time_key=args.time_key, max_len_recent=args.max_len_recent, device=device,
sample_size=args.sample_size, sampling=args.sampling)
with open(f'dataset_{args.data}.pkl', 'wb') as f:
pickle.dump(dataset, f)
print('write dataset to disk')
else:
print('load data')
with open(f'dataset_{args.data}.pkl', 'rb') as f:
dataset = pickle.load(f)
# profiler.stop()
# profiler.print()
# exit()
train_loader = DataLoader(dataset, batch_size=args.batch_size, shuffle=True,
collate_fn=lambda p: collate_fn(p, device=device))
model = Recommender(args.dim, dataset.user_number,
dataset.item_number, device=device)
model = model.to(device)
# todo load model from file
if args.load and os.path.exists(args.save_path):
model.load_state_dict(torch.load(
args.save_path, map_location=device))
return
optimizer = torch.optim.Adam(
model.parameters(), lr=args.lr, weight_decay=args.l2)
n_batch = (dataset.session_number // args.batch_size) + 1
best_recall20 = 0
best_test = ''
log_file = open(PRJ_PATH/'results' /
f'my-{args.data}-ss{args.sample_size}-c{args.c:.1f}-b{args.b1:.1f}-{args.d}.txt', 'w')
for k, v in args.__dict__.items():
print(f'{str(k)}: {str(v)}', file=log_file, flush=True)
print(f'{str(k)}: {str(v)}', flush=True)
patience_idx = 0
#! train
for it in range(args.n_epoch):
print(f'Epoch: {(it + 1)} starts...')
print(f'Epoch: {(it + 1)}', file=log_file, flush=True)
loss_list = []
loss_plist = []
loss_clist = []
idx_batch = 1
for users, session, recent, all_item in train_loader:
model.train()
optimizer.zero_grad()
# filtering samples have empty recent item set
users, session, recent, all_item = mask_no_recent(
users, session, recent, all_item)
items_to_score, map_label = session.unique(return_inverse=True)
# nonzero = items_to_score != 0
# items_to_score = items_to_score[nonzero]
scores, hu, hc, attn = model(
users, session[:, :-1], recent, items_to_score)
loss_predict = model.loss_predict1(
scores, session, map_label)
loss = loss_predict
loss_plist.append(loss_predict.item())
if args.c > 0:
# loss_contrst = model.loss_contrastive(
# all_item, session[:, :-1], hu, hc)
# loss = loss + model.c * loss_contrst
# loss_clist.append(loss_contrst.item())
loss_causal = model.loss_causal(
attn, session[:, :-1], all_item, items_to_score
)
loss = loss + args.c * loss_causal
loss_clist.append(loss_causal.item())
loss_list.append(loss.item())
loss.backward()
optimizer.step()
print(f'\rbatch {idx_batch}/{n_batch} loss: {loss.item():.5f}',
end='') # loss1: {loss_predict.item():.5f} loss2: {loss_contrst.item():.5f}
idx_batch += 1
assert torch.isnan(model.item_embedding.weight).sum() == 0
assert torch.isnan(model.user_embedding.weight).sum() == 0
epoch_loss = np.array(loss_list).mean()
print(
f'\nEpoch: {(it + 1)} total loss: {epoch_loss}, loss1 {np.array(loss_plist).mean()}, loss2 {np.array(loss_clist).mean()}')
print('====================================================================')
#! evaluation
l = [] # 每个测试session,百分之多少的target item在item to predict中
if True: # it > 0 and (it+1)//args.eval_every == 0:
print('Start evaluation...')
print(f'total: {len(dataset.session_ids_test)} sessions to test.')
tester_val = Tester(k_list=[5, 10, 20])
tester_tst = Tester(k_list=[5, 10, 20])
for idx, test_session in tqdm(enumerate(dataset.session_ids_test)):
session_item, users, recent, similarity, items_to_boost = dataset.next_test_session(
test_session, device)
session_context = session_item[:-1].unsqueeze(0)
with torch.no_grad():
scores = model.test_session(users, session_context,
recent, similarity)
scores = scores * args.b2
# scores = scores + float(args.b1) * \
# items_to_boost[np.newaxis, :]
boost_score = float(args.b1) * \
items_to_boost[np.newaxis, :]
# if float(args.b2) > 0: # boost context item neighbors
# items_to_boost2 = dataset.next_test_session2(
# test_session) # context boost item
# boost_score = boost_score + \
# float(args.b2) * items_to_boost2
# boost_score[boost_score > max(float(args.b1), float(args.b2))] = max(
# float(args.b1), float(args.b2))
scores = scores + boost_score
sorted_score_index = scores.argsort(1)[:, ::-1]
predicted_seq = sorted_score_index + 1 # todo 只适用于预测所有item
# predicted_seq = np.arange(1, (len(dataset.items)+1))[
# sorted_score_index]
# predicted_seq = items_to_predict.detach().cpu().numpy()[
# sorted_score_index]
target_seq = session_item[1:].detach().cpu().numpy()
# l.append((np.isin(np.unique(target_seq), np.unique(
# predicted_seq)).sum()/len(np.unique(target_seq))))
# l.extend(items_to_boost[target_seq-1])
# assert (np.isin(target_seq, np.unique(
# predicted_seq)).sum()/len(target_seq)) == 1
# rank = np.nonzero(predicted_seq == target_seq[:, None])[1]
# print(rank) # todo ground label rank
if idx % 2 == 0:
tester_val.evaluate_sequence(
predicted_seq, target_seq, len(target_seq))
else:
tester_tst.evaluate_sequence(
predicted_seq, target_seq, len(target_seq))
score_message, recall5, recall20 = tester_val.get_stats()
# np.savetxt('a.txt', np.array(l)[:, None])
# print(np.mean(l))
print(score_message)
print(score_message, file=log_file, flush=True)
if recall20 > best_recall20:
patience_idx = 0
print('better recall@20!')
print(f'better recall@20!', file=log_file, flush=True)
best_recall20 = recall20
best_test, _, _ = tester_tst.get_stats()
print('Result on test set:')
print(best_test, file=log_file, flush=True)
print(best_test)
else:
patience_idx += 1
print(f'patience {patience_idx}|{args.patience}')
if patience_idx >= args.patience:
print(f'early stop!')
break
print('train finished. \nbest performence:')
print(f'train finished.', file=log_file, flush=True)
print(best_test)
print(best_test, file=log_file, flush=True)
torch.save(model.state_dict(), args.save_path)
print('result saved in: ', PRJ_PATH/'results' /
f'my-{args.data}-ss{args.sample_size}-c{args.c:.1f}-b{args.b1:.1f}-{args.d}.txt')
log_file.close()
def mask_no_recent(user_ids, sess_item, rcnt_item, all_item):
'''
过滤掉recnet_item为空的sample
'''
mask = rcnt_item.sum(1) != 0
if mask.sum() == len(mask):
return user_ids, sess_item, rcnt_item, all_item
# if mask.sum() == 0:
# print('no sample has recent items')
user_ids_ = user_ids[mask]
sess_item_ = sess_item[mask]
sess_len = (sess_item_ != 0).sum(1).max()
sess_item_ = sess_item_[:, -sess_len:]
rcnt_item_ = rcnt_item[mask]
all_item_ = [d for d, m in zip(all_item, mask) if m]
return user_ids_, sess_item_, rcnt_item_, all_item_
class Tester:
def __init__(self, session_length=3, k_list=[5, 10, 20]):
self.k_list = k_list
self.session_length = session_length
self.n_decimals = 4
self.initialize()
def initialize(self):
self.i_count = np.zeros(self.session_length) # [0]*self.session_length
# [[0]*len(self.k) for i in range(self.session_length)]
self.recall = np.zeros((self.session_length, len(self.k_list)))
# [[0]*len(self.k) for i in range(self.session_length)]
self.mrr = np.zeros((self.session_length, len(self.k_list)))
self.ndcg = np.zeros((self.session_length, len(self.k_list)))
def get_rank(self, target, predictions):
for i in range(len(predictions)):
if target == predictions[i]:
return i+1
raise Exception("could not find target in sequence")
def evaluate_sequence(self, predicted_sequence, target_sequence, seq_len):
for i in range(min(self.session_length, seq_len)):
target_item = target_sequence[i]
k_predictions = predicted_sequence[i]
for j in range(len(self.k_list)):
k = self.k_list[j]
if target_item in k_predictions[:k]:
self.recall[i][j] += 1
rank = self.get_rank(target_item, k_predictions[:k])
self.mrr[i][j] += 1.0/rank
self.ndcg[i][j] += 1 / math.log(rank + 1, 2)
self.i_count[i] += 1
def evaluate_batch(self, predictions, targets, sequence_lengths):
for batch_index in range(len(predictions)):
predicted_sequence = predictions[batch_index]
target_sequence = targets[batch_index]
self.evaluate_sequence(
predicted_sequence, target_sequence, sequence_lengths[batch_index])
def get_stats(self):
score_message = "Position\tR@5 \tMRR@5 \tNDCG@5\tR@10 \tMRR@10\tNDCG@10\tR@20 \tMRR@20\tNDCG@20\n"
current_recall = np.zeros(len(self.k_list))
current_mrr = np.zeros(len(self.k_list))
current_ndcg = np.zeros(len(self.k_list))
current_count = 0
recall_k = np.zeros(len(self.k_list))
for i in range(self.session_length):
score_message += "\ni<="+str(i+2)+" \t"
current_count += self.i_count[i]
for j in range(len(self.k_list)):
current_recall[j] += self.recall[i][j]
current_mrr[j] += self.mrr[i][j]
current_ndcg[j] += self.ndcg[i][j]
r = current_recall[j]/current_count
m = current_mrr[j]/current_count
n = current_ndcg[j]/current_count
score_message += str(round(r, self.n_decimals))+'\t'
score_message += str(round(m, self.n_decimals))+'\t'
score_message += str(round(n, self.n_decimals))+'\t'
recall_k[j] = r
recall5 = recall_k[0]
recall20 = recall_k[2]
return score_message, recall5, recall20
if __name__ == '__main__':
main()
# class trainer:
# '''
# Parameters
# -----------
# k : int
# Number of neighboring session to calculate the item scores from. (Default value: 100)
# sample_size : int
# Defines the length of a subset of all training sessions to calculate the nearest neighbors from. (Default value: 500)
# sampling : string
# String to define the sampling method for sessions (recent, random). (default: recent)
# similarity : string
# String to define the method for the similarity calculation (jaccard, cosine, binary, tanimoto). (default: jaccard)
# weighting : string
# Decay function to determine the importance/weight of individual actions in the current session (linear, same, div, log, quadratic). (default: div)
# weighting_score : string
# Decay function to lower the score of candidate items from a neighboring sessions that were selected by less recently clicked items in the current session. (linear, same, div, log, quadratic). (default: div_score)
# weighting_time : boolean
# Experimental function to give less weight to items from older sessions (default: False)
# dwelling_time : boolean
# Experimental function to use the dwelling time for item view actions as a weight in the similarity calculation. (default: False)
# last_n_days : int
# Use only data from the last N days. (default: None)
# last_n_clicks : int
# Use only the last N clicks of the current session when recommending. (default: None)
# extend : bool
# Add evaluated sessions to the maps.
# normalize : bool
# Normalize the scores in the end.
# session_key : string
# Header of the session ID column in the input file. (default: 'SessionId')
# item_key : string
# Header of the item ID column in the input file. (default: 'ItemId')
# time_key : string
# Header of the timestamp column in the input file. (default: 'Time')
# user_key : string
# Header of the user ID column in the input file. (default: 'UserId')
# extend_session_length: int
# extend the current user's session
# extending_mode: string
# how extend the user session (default: lastViewed)
# lastViewed: extend the current user's session with the his/her last viewed items #TODO: now it saves just X last items, and they might be exactly the same: can try as well: save 5 distinctive items
# score_based: higher score: if the items appeared in more previous sessions AND more recently #TODO
# boost_own_sessions: double
# to increase the impact of (give weight more weight to) the sessions which belong to the user. (default: None)
# the value will be added to 1.0. For example for boost_own_sessions=0.2, weight will be 1.2
# past_neighbors: bool
# Include the neighbours of the past user's similar sessions (to the current session) as neighbours (default: False)
# reminders: bool
# Include reminding items in the (main) recommendation list. (default: False)
# remind_strategy: string
# Ranking strategy of the reminding list (recency, session_similarity). (default: recency)
# remind_sessions_num: int
# Number of the last user's sessions that the possible items for reminding are taken from (default: 6)
# reminders_num: int
# length of the reminding list (default: 3)
# remind_mode: string
# The postion of the remining items in recommendation list (top, end). (default: end)
# '''
# def __init__(self, k, sample_size=1000, sampling='random', weighting='div', last_n_days=None, last_n_clicks=None,
# normalize=True, device='cpu',
# session_key='SessionId', item_key='ItemId', time_key='Time', user_key='UserId',
# dim=64, batch_size=128, n_epoch=50, lr=1e-2, l2=1e-5,
# c=0.1, margin=0.2):
# self.k = k
# # ! 注意与原实现不同,当self.sample_size=0的时候,验证阶段只使用当前用户当前context,加模型来预测next item,不再使用knn思想
# self.sample_size = sample_size # 验证时,选择的最大相关context个数,
# self.sampling = sampling # 验证时,采样相关context的策略,包含recent random
# # 验证阶段计算recent items相似度,包括div, same, log, quadratic, linear, 实现见self对应function
# self.weighting = weighting
# self.session_key = session_key # 以下四个key是dataframe的列名
# self.item_key = item_key
# self.time_key = time_key
# self.user_key = user_key
# self.normalize = normalize # 归一化结果
# self.last_n_days = last_n_days
# self.last_n_clicks = last_n_clicks # 当前session只考虑最近多少个click
# # updated while recommending
# self.session = -1 # 验证时,session id
# self.user = -1 # 验证session的user
# self.session_items = [] # 验证session的item列表
# self.recent_items = [] # 验证session的用户的recent item list
# # 验证session的相关用户列表,根据当前session的context中的每个item,寻找其他包含该item的session
# self.relevant_sessions = set()
# self.recent_neighbors = set()
# self.load = True
# # self.load = False
# self.save_path = Path(__file__).parents[2] / 'model.pkl'
# # model parameterss
# self.dim = dim # dim
# self.device = device
# print(f'use device: {device}.')
# self.lr = lr
# self.l2 = l2
# self.n_epoch = n_epoch
# self.batch_size = batch_size
# self.c = c # weight of contrastive loss
# self.margin = margin
# def prepare_for_predict(self, dataset):
# self.session_item_map = dataset.session_item_map
# self.item_session_map = dataset.item_session_map
# self.item_user_map = dataset.item_user_map
# self.user_item_map = dataset.user_item_map
# self.session_user_map = dataset.session_user_map
# self.user_session_map = dataset.user_session_map
# self.session_time = dataset.session_time
# self.session_recent_map = dataset.session_recent_map
# self.recent_session_map = dataset.recent_session_map
# self.user_recent_map = dataset.user_recent_map
# self.max_len_recent = dataset.max_len_recent
# self.user2id = dataset.user2id
# self.item2id = dataset.item2id
# self.id2item = dataset.id2item
# def predict_next(self, session_id, input_item_id, input_user_id, predict_for_item_ids=None, timestamp=0):
# """
# Gives predicton scores for a selected set of items on how likely they be the next item in the session.
# Parameters
# --------
# session_id : int or string
# The session IDs of the event.
# input_item_id : int or string
# The item ID of the event. Must be in the set of item IDs of the training set.
# predict_for_item_ids : 1D array
# IDs of items for which the network should give prediction scores. Every ID must be in the set of item IDs of the training set.
# Returns
# --------
# out : pandas.Series
# Prediction scores for selected items on how likely to be the next item of this session. Indexed by the item IDs.
# """
# device = 'cpu' # 验证
# predict_for_item_ids = [self.id2item[i] for i in range(
# 1, len(self.item2id)+1)] # 将待预测的item按照item embedding顺序重排列
# if self.session != session_id: # new session
# # add evaluated sessions to the maps.
# # ! 问题:上一个session的item并不完整,因为predict_next并未扫描到每个session最后一个item
# item_set = set(self.session_items)
# if len(item_set) > 0:
# self.session_item_map[self.session] = item_set
# for item in item_set:
# map_is = self.item_session_map.get(item)
# if map_is is None:
# map_is = set()
# self.item_session_map.update({item: map_is})
# map_is.add(self.session)
# ts = time.time()
# self.session_time.update({self.session: ts})
# self.session_user_map.update({self.session: self.user})
# # 先更新recent map
# self.session_recent_map.update(
# {self.session: self.recent_items.copy()}) # session id到包含的item列表的映射
# for item in set(self.recent_items): # 构造recent记录中item到session id的映射
# if item != 0:
# if item in self.recent_session_map:
# self.recent_session_map[item].add(self.session)
# else:
# self.recent_session_map[item] = {self.session}
# # 再更新recent
# self.recent_items = self.user_recent_map.get(self.user,
# []) # 将上一个处理的session内容更新到recent_items,对应上一个处理的session和user
# self.recent_items.extend(self.session_items)
# if self.max_len_recent > 0:
# self.recent_items = self.recent_items[-self.max_len_recent:]
# self.user_recent_map.update(
# {self.user: self.recent_items.copy()}) # user id到recent item的映射
# # 新验证session, 初始化验证session的信息
# self.last_ts = -1
# self.session = session_id
# self.session_items = list() # 验证session的所有item的list
# self.user = self.user2id[input_user_id]
# self.recent_items = self.user_recent_map[self.user]
# if self.sample_size > 0:
# self.recent_neighbors = self.recent_neighbor_sessions(
# self.recent_items)
# self.relevant_sessions = set()
# item = self.item2id[input_item_id]
# # 将扫描到的item加入到当前session的item列表,input_item_id是context最近的一个item
# self.session_items.append(item)
# # 只考虑最后n个click
# self.session_items = self.session_items if self.last_n_clicks is None else self.session_items[
# -self.last_n_clicks:]
# sess_item = torch.LongTensor(
# self.session_items).unsqueeze(0).to(device)
# if self.sample_size > 0:
# # 根据当前session的context,查找可能相关的其它用户的context:对于context中的每个item,查找包含此item的session
# neighbors = self.find_neighbors(item, session_id, self.user)
# # 同时考虑根据context查找的和根据recent items查找的neighbors
# neighbors = neighbors | self.recent_neighbors
# # 计算recent items相似度,取相似度最高的k个
# _, similarities = self.calc_similarity(
# self.recent_items, neighbors)
# sorted_idx = np.argsort(similarities)
# similarities = similarities[sorted_idx[-self.k:]] # 取最相似的k个
# # similarities = np.append(similarities, similarities.sum()) # todo,暂定当前用户的预测结果权重占一半
# similarities = similarities / similarities.sum()
# rcnt_items = np.array([self.session_recent_map[sess_id] for sess_id in neighbors])[
# sorted_idx[-self.k:]] # 过滤掉超出top k的
# # rcnt_items = np.append(rcnt_items, self.recent_items).reshape(-1, rcnt_items.shape[1]) # 补充当前用户
# rcnt_items = torch.LongTensor(rcnt_items).to(device)
# # 构造输入,user id和recent items使用其他用户的,当前session context使用当前用户的
# user_ids = np.array([self.session_user_map[sess_id] for sess_id in neighbors])[
# sorted_idx[-self.k:]] # 过滤掉超出top k的
# # user_ids = np.append(user_ids, self.user) # 补充当前用户 预测结果包括当前用户的recent items+当前session context
# user_ids = torch.LongTensor(user_ids).to(device)
# else: # 若sample_size==0, 只使用当前context和recent items,加上模型来预测
# user_ids = torch.LongTensor([self.user]).to(device)
# rcnt_items = torch.LongTensor([self.recent_items]).to(device)
# similarities = np.ones(1)
# neighbors = set()
# if len(neighbors) > 0:
# items_to_predict = set()
# for s in neighbors:
# items_to_predict = items_to_predict | set(
# self.session_item_map[s])
# items_to_predict = torch.LongTensor(
# list(items_to_predict)).to(device)
# else:
# items_to_predict = torch.LongTensor(
# list(map(self.item2id.get, predict_for_item_ids))).to(device)
# # 模型预测查找到的相似session context下,不同recent item下在同一context下用户的选择
# self.model.to(device) # faster than cal with gpu
# self.model.eval()
# with torch.no_grad():
# scores, _, _ = self.model(
# user_ids, sess_item, rcnt_items, items_to_predict)
# # (num_neighbor+1)*num_item2predict, pytorch tensor to numpy list
# scores = scores.detach().cpu().numpy()
# # 根据相似度聚合,若sample_size==0, 则similarities = [1]
# scores = (scores * similarities[:, np.newaxis]).sum(0)
# # scores = self.score_items(neighbors, items, timestamp) # 返回一个item id:score的dict
# # Create things in the format ..
# predictions = np.zeros(len(self.item2id))
# for item, score in zip(items_to_predict, scores):
# predictions[(item - 1)] = score
# if self.normalize: # scale to 0-1
# predictions = predictions - predictions.max()
# predictions = np.exp(predictions)
# # mask = np.in1d(predict_for_item_ids, list(scores.keys()))
# # predict_for_items = predict_for_item_ids[mask]
# # values = [scores[x] for x in predict_for_items]
# # predictions[mask] = values
# series = pd.Series(data=predictions, index=predict_for_item_ids)
# return series
# def mask_no_recent(self, user_ids, sess_item, rcnt_item, all_item):
# '''
# 过滤掉recnet_item为空的sample
# '''
# mask = rcnt_item.sum(1) != 0
# if mask.sum() == len(mask):
# return user_ids, sess_item, rcnt_item, all_item
# # if mask.sum() == 0:
# # print('no sample has recent items')
# user_ids_ = user_ids[mask]
# sess_item_ = sess_item[mask]
# rcnt_item_ = rcnt_item[mask]
# all_item_ = [d for d, m in zip(all_item, mask) if m]
# return user_ids_, sess_item_, rcnt_item_, all_item_
# def item_pop(self, sessions):
# '''
# Returns a dict(item,score) of the item popularity for the given list of sessions (only a set of ids)
# Parameters
# --------
# sessions: set
# Returns
# --------
# out : dict
# '''
# result = dict()
# max_pop = 0
# for session, weight in sessions:
# items = self.items_for_session(session)
# for item in items:
# count = result.get(item)
# if count is None:
# result.update({item: 1})
# else:
# result.update({item: count + 1})
# if (result.get(item) > max_pop):
# max_pop = result.get(item)
# for key in result:
# result.update({key: (result[key] / max_pop)})
# return result
# def jaccard(self, first, second):
# '''
# Calculates the jaccard index for two sessions
# Parameters
# --------
# first: Id of a session
# second: Id of a session
# Returns
# --------
# out : float value
# '''
# sc = time.clock()
# intersection = len(first & second)
# union = len(first | second)
# res = intersection / union
# self.sim_time += (time.clock() - sc)
# return res
# def cosine(self, first, second):
# '''
# Calculates the cosine similarity for two sessions
# Parameters
# --------
# first: Id of a session
# second: Id of a session
# Returns
# --------
# out : float value
# '''
# li = len(first & second)
# la = len(first)
# lb = len(second)
# result = li / sqrt(la) * sqrt(lb)
# return result
# def tanimoto(self, first, second):
# '''
# Calculates the cosine tanimoto similarity for two sessions
# Parameters
# --------
# first: Id of a session
# second: Id of a session
# Returns
# --------
# out : float value
# '''
# li = len(first & second)
# la = len(first)
# lb = len(second)
# result = li / (la + lb - li)
# return result
# def binary(self, first, second):
# '''
# Calculates the ? for 2 sessions
# Parameters
# --------
# first: Id of a session
# second: Id of a session
# Returns
# --------
# out : float value
# '''
# a = len(first & second)
# b = len(first)
# c = len(second)
# result = (2 * a) / ((2 * a) + b + c)
# return result
# def vec(self, first, second, map):
# '''
# Calculates the ? for 2 sessions
# Parameters
# --------
# first: Id of a session
# second: Id of a session
# Returns
# --------
# out : float value
# '''
# a = first & second
# sum = 0
# for i in a:
# sum += map[i]
# result = sum / len(map)
# return result
# def items_for_session(self, session):
# '''
# Returns all items in the session
# Parameters
# --------
# session: Id of a session
# Returns
# --------
# out : set
# '''
# return self.session_item_map.get(session)
# def vec_for_session(self, session):
# '''
# Returns all items in the session
# Parameters
# --------
# session: Id of a session
# Returns
# --------
# out : set
# '''
# return self.session_vec_map.get(session)
# def sessions_for_item(self, item_id):
# '''
# Returns all session for an item
# Parameters
# --------
# item: Id of the item session
# Returns
# --------
# out : set
# '''
# return self.item_session_map.get(item_id) if item_id in self.item_session_map else set()
# def most_recent_sessions(self, sessions, number):
# '''
# Find the most recent sessions in the given set
# Parameters
# --------
# sessions: set of session ids
# Returns
# --------
# out : set
# '''
# sample = set()
# tuples = list()
# for session in sessions:
# time = self.session_time.get(session) # ! ?这里不应该是
# if time is None:
# print(' EMPTY TIMESTAMP!! ', session)
# tuples.append((session, time))
# tuples = sorted(tuples, key=itemgetter(1), reverse=True)
# # print 'sorted list ', sortedList
# cnt = 0
# for element in tuples:
# cnt = cnt + 1
# if cnt > number:
# break
# sample.add(element[0])
# # print 'returning sample of size ', len(sample)
# return sample
# def possible_neighbor_sessions(self, input_item_id, session_id, user_id):
# '''
# Find a set of session to later on find neighbors in.
# A self.sample_size of 0 uses all sessions in which any item of the current session appears.
# self.sampling can be performed with the options "recent" or "random".
# "recent" selects the self.sample_size most recent sessions while "random" just choses randomly.
# Parameters
# --------
# sessions: set of session ids
# Returns
# --------
# out : set
# '''
# # add relevant sessions for the current item
# self.relevant_sessions = self.relevant_sessions | self.sessions_for_item(
# input_item_id)
# if self.sample_size == 0: # use all session as possible neighbors
# # print('!!!!! runnig KNN without a sample size (check config)')
# possible_neighbors = self.relevant_sessions
# else: # sample some sessions
# if len(self.relevant_sessions) > self.sample_size:
# if self.sampling == 'recent':
# sample = self.most_recent_sessions(
# self.relevant_sessions, self.sample_size)
# elif self.sampling == 'random':
# sample = random.sample(
# self.relevant_sessions, self.sample_size)
# else:
# sample = self.relevant_sessions[:self.sample_size]
# possible_neighbors = sample
# else:
# possible_neighbors = self.relevant_sessions
# return possible_neighbors
# def recent_neighbor_sessions(self, recent_items):
# '''
# 根据recent items查找neighbor sessions
# 给定当前session的recent列表
# 对其中的每个item,查询recent列表中包含该item的session的id,依赖self.session_recent_map
# '''
# possible_neighbors = set()
# for item in set(recent_items):
# possible_neighbors = possible_neighbors | self.recent_session_map.get(
# item, set())
# if self.sample_size == 0: # use all session as possible neighbors
# # print('!!!!! runnig KNN without a sample size (check config)')
# result = possible_neighbors
# else: # sample some sessions
# if len(possible_neighbors) > self.sample_size:
# if self.sampling == 'recent':
# sample = self.most_recent_sessions(
# possible_neighbors, self.sample_size)
# elif self.sampling == 'random':
# sample = random.sample(
# possible_neighbors, self.sample_size)
# else:
# sample = possible_neighbors[:self.sample_size]
# result = sample
# else:
# result = possible_neighbors
# return result
# def calc_similarity(self, recent_items, sessions):
# '''
# Calculates the configured similarity for the items in recent_items and each session in sessions.
# Parameters
# --------
# recent_items: set of item ids
# sessions: list of session ids
# Returns
# --------
# out : list of tuple (session_id,similarity)
# '''
# pos_map = {} # 计算当前session的item权重,根据位置
# length = len(recent_items)
# count = 1
# for item in recent_items:
# if self.weighting is not None:
# pos_map[item] = getattr(self, self.weighting)(count, length)
# count += 1
# else:
# pos_map[item] = 1
# # if self.dwelling_time:
# # dt = dwelling_times.copy()
# # dt.append(0)
# # dt = pd.Series(dt, index=session_items)
# # dt = dt / dt.max()
# # # dt[session_items[-1]] = dt.mean() if len(session_items) > 1 else 1
# # dt[session_items[-1]] = 1
# # # print(dt)
# # for i in range(len(dt)):
# # pos_map[session_items[i]] *= dt.iloc[i]
# # print(pos_map)
# # if self.idf_weighting_session: # 未使用
# # max = -1
# # for item in session_items:
# # pos_map[item] = self.idf[item] if item in self.idf else 0
# # if pos_map[item] > max:
# # max = pos_map[item]
# # for item in session_items:
# # pos_map[item] = pos_map[item] / max
# # print 'nb of sessions to test ', len(sessionsToTest), ' metric: ', self.metric
# items = set(recent_items)
# neighbors = []
# similarities = []
# cnt = 0
# for session in sessions: # 对于每个可能的相似session
# cnt = cnt + 1
# # get items of the session, look up the cache first
# # n_items = self.items_for_session(session) # 相似session的item set
# n_items = set(self.session_recent_map[session])
# # sts = self.session_time[session]
# # dot product
# # 计算session相似度,内积,但考虑每个共同item根据其在当前session中的位置
# similarity = self.vec(items, n_items, pos_map)
# if similarity > 0:
# # if self.weighting_time: # 未使用
# # diff = timestamp - sts
# # days = round(diff / 60 / 60 / 24)
# # decay = pow(7 / 8, days)
# # similarity *= decay