/
index.html
5371 lines (5007 loc) · 259 KB
/
index.html
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
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width, initial-scale=1, minimum-scale=1" />
<meta name="generator" content="pdoc 0.9.2" />
<title>ktrain.text API documentation</title>
<meta name="description" content="" />
<link rel="preload stylesheet" as="style" href="https://cdnjs.cloudflare.com/ajax/libs/10up-sanitize.css/11.0.1/sanitize.min.css" integrity="sha256-PK9q560IAAa6WVRRh76LtCaI8pjTJ2z11v0miyNNjrs=" crossorigin>
<link rel="preload stylesheet" as="style" href="https://cdnjs.cloudflare.com/ajax/libs/10up-sanitize.css/11.0.1/typography.min.css" integrity="sha256-7l/o7C8jubJiy74VsKTidCy1yBkRtiUGbVkYBylBqUg=" crossorigin>
<link rel="stylesheet preload" as="style" href="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/10.1.1/styles/github.min.css" crossorigin>
<style>:root{--highlight-color:#fe9}.flex{display:flex !important}body{line-height:1.5em}#content{padding:20px}#sidebar{padding:30px;overflow:hidden}#sidebar > *:last-child{margin-bottom:2cm}.http-server-breadcrumbs{font-size:130%;margin:0 0 15px 0}#footer{font-size:.75em;padding:5px 30px;border-top:1px solid #ddd;text-align:right}#footer p{margin:0 0 0 1em;display:inline-block}#footer p:last-child{margin-right:30px}h1,h2,h3,h4,h5{font-weight:300}h1{font-size:2.5em;line-height:1.1em}h2{font-size:1.75em;margin:1em 0 .50em 0}h3{font-size:1.4em;margin:25px 0 10px 0}h4{margin:0;font-size:105%}h1:target,h2:target,h3:target,h4:target,h5:target,h6:target{background:var(--highlight-color);padding:.2em 0}a{color:#058;text-decoration:none;transition:color .3s ease-in-out}a:hover{color:#e82}.title code{font-weight:bold}h2[id^="header-"]{margin-top:2em}.ident{color:#900}pre code{background:#f8f8f8;font-size:.8em;line-height:1.4em}code{background:#f2f2f1;padding:1px 4px;overflow-wrap:break-word}h1 code{background:transparent}pre{background:#f8f8f8;border:0;border-top:1px solid #ccc;border-bottom:1px solid #ccc;margin:1em 0;padding:1ex}#http-server-module-list{display:flex;flex-flow:column}#http-server-module-list div{display:flex}#http-server-module-list dt{min-width:10%}#http-server-module-list p{margin-top:0}.toc ul,#index{list-style-type:none;margin:0;padding:0}#index code{background:transparent}#index h3{border-bottom:1px solid #ddd}#index ul{padding:0}#index h4{margin-top:.6em;font-weight:bold}@media (min-width:200ex){#index .two-column{column-count:2}}@media (min-width:300ex){#index .two-column{column-count:3}}dl{margin-bottom:2em}dl dl:last-child{margin-bottom:4em}dd{margin:0 0 1em 3em}#header-classes + dl > dd{margin-bottom:3em}dd dd{margin-left:2em}dd p{margin:10px 0}.name{background:#eee;font-weight:bold;font-size:.85em;padding:5px 10px;display:inline-block;min-width:40%}.name:hover{background:#e0e0e0}dt:target .name{background:var(--highlight-color)}.name > span:first-child{white-space:nowrap}.name.class > span:nth-child(2){margin-left:.4em}.inherited{color:#999;border-left:5px solid #eee;padding-left:1em}.inheritance em{font-style:normal;font-weight:bold}.desc h2{font-weight:400;font-size:1.25em}.desc h3{font-size:1em}.desc dt code{background:inherit}.source summary,.git-link-div{color:#666;text-align:right;font-weight:400;font-size:.8em;text-transform:uppercase}.source summary > *{white-space:nowrap;cursor:pointer}.git-link{color:inherit;margin-left:1em}.source pre{max-height:500px;overflow:auto;margin:0}.source pre code{font-size:12px;overflow:visible}.hlist{list-style:none}.hlist li{display:inline}.hlist li:after{content:',\2002'}.hlist li:last-child:after{content:none}.hlist .hlist{display:inline;padding-left:1em}img{max-width:100%}td{padding:0 .5em}.admonition{padding:.1em .5em;margin-bottom:1em}.admonition-title{font-weight:bold}.admonition.note,.admonition.info,.admonition.important{background:#aef}.admonition.todo,.admonition.versionadded,.admonition.tip,.admonition.hint{background:#dfd}.admonition.warning,.admonition.versionchanged,.admonition.deprecated{background:#fd4}.admonition.error,.admonition.danger,.admonition.caution{background:lightpink}</style>
<style media="screen and (min-width: 700px)">@media screen and (min-width:700px){#sidebar{width:30%;height:100vh;overflow:auto;position:sticky;top:0}#content{width:70%;max-width:100ch;padding:3em 4em;border-left:1px solid #ddd}pre code{font-size:1em}.item .name{font-size:1em}main{display:flex;flex-direction:row-reverse;justify-content:flex-end}.toc ul ul,#index ul{padding-left:1.5em}.toc > ul > li{margin-top:.5em}}</style>
<style media="print">@media print{#sidebar h1{page-break-before:always}.source{display:none}}@media print{*{background:transparent !important;color:#000 !important;box-shadow:none !important;text-shadow:none !important}a[href]:after{content:" (" attr(href) ")";font-size:90%}a[href][title]:after{content:none}abbr[title]:after{content:" (" attr(title) ")"}.ir a:after,a[href^="javascript:"]:after,a[href^="#"]:after{content:""}pre,blockquote{border:1px solid #999;page-break-inside:avoid}thead{display:table-header-group}tr,img{page-break-inside:avoid}img{max-width:100% !important}@page{margin:0.5cm}p,h2,h3{orphans:3;widows:3}h1,h2,h3,h4,h5,h6{page-break-after:avoid}}</style>
<script defer src="https://cdnjs.cloudflare.com/ajax/libs/highlight.js/10.1.1/highlight.min.js" integrity="sha256-Uv3H6lx7dJmRfRvH8TH6kJD1TSK1aFcwgx+mdg3epi8=" crossorigin></script>
<script>window.addEventListener('DOMContentLoaded', () => hljs.initHighlighting())</script>
</head>
<body>
<main>
<article id="content">
<header>
<h1 class="title">Module <code>ktrain.text</code></h1>
</header>
<section id="section-intro">
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">from .models import print_text_classifiers, print_text_regression_models, text_classifier, text_regression_model
from .data import texts_from_folder, texts_from_csv, texts_from_df, texts_from_array
from .ner.data import entities_from_gmb, entities_from_conll2003, entities_from_txt, entities_from_df, entities_from_array
from .ner.models import sequence_tagger, print_sequence_taggers
from .eda import get_topic_model
from .textutils import extract_filenames, load_text_files, filter_by_id
from .preprocessor import Transformer, TransformerEmbedding
from .summarization import TransformerSummarizer
from .zsl import ZeroShotClassifier
from .translation import EnglishTranslator, Translator
from . import shallownlp
from .qa import SimpleQA
from . import textutils
import pickle
__all__ = [
'text_classifier', 'text_regression_model',
'print_text_classifiers', 'print_text_regression_models',
'texts_from_folder', 'texts_from_csv', 'texts_from_df', 'texts_from_array',
'entities_from_gmb',
'entities_from_conll2003',
'entities_from_txt',
'entities_from_array',
'entities_from_df',
'sequence_tagger',
'print_sequence_taggers',
'get_topic_model',
'Transformer',
'TransformerEmbedding',
'shallownlp',
'TransformerSummarizer',
'ZeroShotClassifier',
'EnglishTranslator',
'Translator',
'SimpleQA',
'extract_filenames',
'load_text_files',
]
def load_topic_model(fname):
"""
Load saved TopicModel object
Args:
fname(str): base filename for all saved files
"""
with open(fname+'.tm_vect', 'rb') as f:
vectorizer = pickle.load(f)
with open(fname+'.tm_model', 'rb') as f:
model = pickle.load(f)
with open(fname+'.tm_params', 'rb') as f:
params = pickle.load(f)
tm = get_topic_model(n_topics=params['n_topics'],
n_features = params['n_features'],
verbose = params['verbose'])
tm.model = model
tm.vectorizer = vectorizer
return tm
seqlen_stats = Transformer.seqlen_stats</code></pre>
</details>
</section>
<section>
<h2 class="section-title" id="header-submodules">Sub-modules</h2>
<dl>
<dt><code class="name"><a title="ktrain.text.data" href="data.html">ktrain.text.data</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.eda" href="eda.html">ktrain.text.eda</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.learner" href="learner.html">ktrain.text.learner</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.models" href="models.html">ktrain.text.models</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.ner" href="ner/index.html">ktrain.text.ner</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.predictor" href="predictor.html">ktrain.text.predictor</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.preprocessor" href="preprocessor.html">ktrain.text.preprocessor</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.qa" href="qa/index.html">ktrain.text.qa</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.shallownlp" href="shallownlp/index.html">ktrain.text.shallownlp</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.summarization" href="summarization/index.html">ktrain.text.summarization</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.textutils" href="textutils.html">ktrain.text.textutils</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.translation" href="translation/index.html">ktrain.text.translation</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
<dt><code class="name"><a title="ktrain.text.zsl" href="zsl/index.html">ktrain.text.zsl</a></code></dt>
<dd>
<div class="desc"></div>
</dd>
</dl>
</section>
<section>
</section>
<section>
<h2 class="section-title" id="header-functions">Functions</h2>
<dl>
<dt id="ktrain.text.entities_from_array"><code class="name flex">
<span>def <span class="ident">entities_from_array</span></span>(<span>x_train, y_train, x_test=None, y_test=None, use_char=False, val_pct=0.1, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><p>Load entities from arrays</p>
<h2 id="args">Args</h2>
<p>x_train(list): list of list of entity tokens for training
Example: x_train = [['Hello', 'world'], ['Hello', 'Cher'], ['I', 'love', 'Chicago']]
y_train(list): list of list of tokens representing entity labels
Example:
y_train = [['O', 'O'], ['O', 'B-PER'], ['O', 'O', 'B-LOC']]
x_test(list): list of list of entity tokens for validation
Example: x_train = [['Hello', 'world'], ['Hello', 'Cher'], ['I', 'love', 'Chicago']]
y_test(list): list of list of tokens representing entity labels
Example:
y_train = [['O', 'O'], ['O', 'B-PER'], ['O', 'O', 'B-LOC']]
use_char(bool):
If True, data will be preprocessed to use character embeddings
in addition to word embeddings
val_pct(float):
percentage of training to use for validation if no validation data is supplied
verbose (boolean): verbosity</p></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def entities_from_array(x_train, y_train,
x_test=None, y_test=None,
use_char=False,
val_pct=0.1,
verbose=1):
"""
Load entities from arrays
Args:
x_train(list): list of list of entity tokens for training
Example: x_train = [['Hello', 'world'], ['Hello', 'Cher'], ['I', 'love', 'Chicago']]
y_train(list): list of list of tokens representing entity labels
Example: y_train = [['O', 'O'], ['O', 'B-PER'], ['O', 'O', 'B-LOC']]
x_test(list): list of list of entity tokens for validation
Example: x_train = [['Hello', 'world'], ['Hello', 'Cher'], ['I', 'love', 'Chicago']]
y_test(list): list of list of tokens representing entity labels
Example: y_train = [['O', 'O'], ['O', 'B-PER'], ['O', 'O', 'B-LOC']]
use_char(bool): If True, data will be preprocessed to use character embeddings in addition to word embeddings
val_pct(float): percentage of training to use for validation if no validation data is supplied
verbose (boolean): verbosity
"""
# TODO: converting to df to use entities_from_df - needs to be refactored
train_df = pp.array_to_df(x_train, y_train)
val_df = None
if x_test is not None and y_test is not None:
val_df = pp.array_to_df(x_test, y_test)
if verbose:
print('training data sample:')
print(train_df.head())
if val_df is not None:
print('validation data sample:')
print(val_df.head())
return entities_from_df(train_df, val_df=val_df, val_pct=val_pct,
use_char=use_char, verbose=verbose)</code></pre>
</details>
</dd>
<dt id="ktrain.text.entities_from_conll2003"><code class="name flex">
<span>def <span class="ident">entities_from_conll2003</span></span>(<span>train_filepath, val_filepath=None, use_char=False, encoding=None, val_pct=0.1, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><p>Loads sequence-labeled data from a file in CoNLL2003 format.</p></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def entities_from_conll2003(train_filepath,
val_filepath=None,
use_char=False,
encoding=None,
val_pct=0.1, verbose=1):
"""
Loads sequence-labeled data from a file in CoNLL2003 format.
"""
return entities_from_txt(train_filepath=train_filepath,
val_filepath=val_filepath,
use_char=use_char,
data_format='conll2003',
encoding=encoding,
val_pct=val_pct, verbose=verbose)</code></pre>
</details>
</dd>
<dt id="ktrain.text.entities_from_df"><code class="name flex">
<span>def <span class="ident">entities_from_df</span></span>(<span>train_df, val_df=None, word_column='Word', tag_column='Tag', sentence_column='SentenceID', use_char=False, val_pct=0.1, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><p>Load entities from pandas DataFrame</p>
<h2 id="args">Args</h2>
<dl>
<dt>train_df(pd.DataFrame): training data</dt>
<dt>val_df(pdf.DataFrame): validation data</dt>
<dt>word_column(str): name of column containing the text</dt>
<dt>tag_column(str): name of column containing lael</dt>
<dt>sentence_column(str): name of column containing Sentence IDs</dt>
<dt>use_char(bool):
If True, data will be preprocessed to use character embeddings
in addition to word embeddings</dt>
<dt><strong><code>verbose</code></strong> : <code>boolean</code></dt>
<dd>verbosity</dd>
</dl></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def entities_from_df(train_df,
val_df=None,
word_column=WORD_COL,
tag_column=TAG_COL,
sentence_column=SENT_COL,
use_char=False,
val_pct=0.1, verbose=1):
"""
Load entities from pandas DataFrame
Args:
train_df(pd.DataFrame): training data
val_df(pdf.DataFrame): validation data
word_column(str): name of column containing the text
tag_column(str): name of column containing lael
sentence_column(str): name of column containing Sentence IDs
use_char(bool): If True, data will be preprocessed to use character embeddings in addition to word embeddings
verbose (boolean): verbosity
"""
# process dataframe and instantiate NERPreprocessor
x, y = pp.process_df(train_df,
word_column=word_column,
tag_column=tag_column,
sentence_column=sentence_column,
verbose=verbose)
# get validation set
if val_df is None:
x_train, x_valid, y_train, y_valid = train_test_split(x, y, test_size=val_pct)
else:
x_train, y_train = x, y
(x_valid, y_valid) = pp.process_df(val_df,
word_column=word_column,
tag_column=tag_column,
sentence_column=sentence_column,
verbose=0)
# preprocess and convert to generator
p = IndexTransformer(use_char=use_char)
preproc = NERPreprocessor(p)
preproc.fit(x_train, y_train)
trn = pp.NERSequence(x_train, y_train, batch_size=U.DEFAULT_BS, p=p)
val = pp.NERSequence(x_valid, y_valid, batch_size=U.DEFAULT_BS, p=p)
return ( trn, val, preproc)</code></pre>
</details>
</dd>
<dt id="ktrain.text.entities_from_gmb"><code class="name flex">
<span>def <span class="ident">entities_from_gmb</span></span>(<span>train_filepath, val_filepath=None, use_char=False, word_column='Word', tag_column='Tag', sentence_column='SentenceID', encoding=None, val_pct=0.1, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><p>Loads sequence-labeled data from text file in the
Groningen
Meaning Bank
(GMB) format.</p></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def entities_from_gmb(train_filepath,
val_filepath=None,
use_char=False,
word_column=WORD_COL,
tag_column=TAG_COL,
sentence_column=SENT_COL,
encoding=None,
val_pct=0.1, verbose=1):
"""
Loads sequence-labeled data from text file in the Groningen
Meaning Bank (GMB) format.
"""
return entities_from_txt(train_filepath=train_filepath,
val_filepath=val_filepath,
use_char=use_char,
word_column=word_column,
tag_column=tag_column,
sentence_column=sentence_column,
data_format='gmb',
encoding=encoding,
val_pct=val_pct, verbose=verbose)</code></pre>
</details>
</dd>
<dt id="ktrain.text.entities_from_txt"><code class="name flex">
<span>def <span class="ident">entities_from_txt</span></span>(<span>train_filepath, val_filepath=None, use_char=False, word_column='Word', tag_column='Tag', sentence_column='SentenceID', data_format='conll2003', encoding=None, val_pct=0.1, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><p>Loads sequence-labeled data from comma or tab-delmited text file.
Format of file is either the CoNLL2003 format or Groningen Meaning
Bank (GMB) format - specified with data_format parameter.</p>
<p>In both formats, each word appars on a separate line along with
its associated tag (or label).<br>
The last item on each line should be the tag or label assigned to word.</p>
<p>In the CoNLL2003 format, there is an empty line after
each sentence.
In the GMB format, sentences are deliniated
with a third column denoting the Sentence ID.</p>
<p>More information on CoNLL2003 format:
<a href="https://www.aclweb.org/anthology/W03-0419">https://www.aclweb.org/anthology/W03-0419</a></p>
<p>CoNLL Example (each column is typically separated by space or tab)
and
no column headings:</p>
<p>Paul
B-PER
Newman
I-PER
is
O
a
O
great
O
actor
O
!
O</p>
<p>More information on GMB format:
Refer to ner_dataset.csv on Kaggle here:
<a href="https://www.kaggle.com/abhinavwalia95/entity-annotated-corpus/version/2">https://www.kaggle.com/abhinavwalia95/entity-annotated-corpus/version/2</a></p>
<p>GMB example (each column separated by comma or tab)
with column headings:</p>
<p>SentenceID
Word
Tag
<br>
1
Paul
B-PER
1
Newman
I-PER
1
is
O
1
a
O
1
great
O
1
actor
O
1
!
O</p>
<h2 id="args">Args</h2>
<dl>
<dt>train_filepath(str): file path to training CSV</dt>
<dt><strong><code>val_filepath</code></strong> : <code>str</code></dt>
<dd>file path to validation dataset</dd>
<dt>use_char(bool):
If True, data will be preprocessed to use character embeddings in addition to word embeddings</dt>
<dt>word_column(str): name of column containing the text</dt>
<dt>tag_column(str): name of column containing lael</dt>
<dt>sentence_column(str): name of column containing Sentence IDs</dt>
<dt>data_format(str): one of colnll2003 or gmb</dt>
<dt>word_column, tag_column, and sentence_column</dt>
<dt>ignored if 'conll2003'</dt>
<dt>encoding(str): the encoding to use.
If None, encoding is discovered automatically</dt>
<dt>val_pct(float): Proportion of training to use for validation.</dt>
<dt><strong><code>verbose</code></strong> : <code>boolean</code></dt>
<dd>verbosity</dd>
</dl></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def entities_from_txt(train_filepath,
val_filepath=None,
use_char=False,
word_column=WORD_COL,
tag_column=TAG_COL,
sentence_column=SENT_COL,
data_format='conll2003',
encoding=None,
val_pct=0.1, verbose=1):
"""
Loads sequence-labeled data from comma or tab-delmited text file.
Format of file is either the CoNLL2003 format or Groningen Meaning
Bank (GMB) format - specified with data_format parameter.
In both formats, each word appars on a separate line along with
its associated tag (or label).
The last item on each line should be the tag or label assigned to word.
In the CoNLL2003 format, there is an empty line after
each sentence. In the GMB format, sentences are deliniated
with a third column denoting the Sentence ID.
More information on CoNLL2003 format:
https://www.aclweb.org/anthology/W03-0419
CoNLL Example (each column is typically separated by space or tab)
and no column headings:
Paul B-PER
Newman I-PER
is O
a O
great O
actor O
! O
More information on GMB format:
Refer to ner_dataset.csv on Kaggle here:
https://www.kaggle.com/abhinavwalia95/entity-annotated-corpus/version/2
GMB example (each column separated by comma or tab)
with column headings:
SentenceID Word Tag
1 Paul B-PER
1 Newman I-PER
1 is O
1 a O
1 great O
1 actor O
1 ! O
Args:
train_filepath(str): file path to training CSV
val_filepath (str): file path to validation dataset
use_char(bool): If True, data will be preprocessed to use character embeddings in addition to word embeddings
word_column(str): name of column containing the text
tag_column(str): name of column containing lael
sentence_column(str): name of column containing Sentence IDs
data_format(str): one of colnll2003 or gmb
word_column, tag_column, and sentence_column
ignored if 'conll2003'
encoding(str): the encoding to use. If None, encoding is discovered automatically
val_pct(float): Proportion of training to use for validation.
verbose (boolean): verbosity
"""
# set dataframe converter
if data_format == 'gmb':
data_to_df = pp.gmb_to_df
else:
data_to_df = pp.conll2003_to_df
word_column, tag_column, sentence_column = WORD_COL, TAG_COL, SENT_COL
# detect encoding
if encoding is None:
with open(train_filepath, 'rb') as f:
encoding = TU.detect_encoding(f.read())
U.vprint('detected encoding: %s (if wrong, set manually)' % (encoding), verbose=verbose)
# create dataframe
train_df = data_to_df(train_filepath, encoding=encoding)
val_df = None if val_filepath is None else data_to_df(val_filepath, encoding=encoding)
return entities_from_df(train_df,
val_df=val_df,
word_column=word_column,
tag_column=tag_column,
sentence_column=sentence_column,
use_char=use_char,
val_pct=val_pct, verbose=verbose)</code></pre>
</details>
</dd>
<dt id="ktrain.text.extract_filenames"><code class="name flex">
<span>def <span class="ident">extract_filenames</span></span>(<span>corpus_path, follow_links=False)</span>
</code></dt>
<dd>
<div class="desc"></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def extract_filenames(corpus_path, follow_links=False):
if os.listdir(corpus_path) == []:
raise ValueError("%s: path is empty" % corpus_path)
walk = os.walk
for root, dirs, filenames in walk(corpus_path, followlinks=follow_links):
for filename in filenames:
try:
yield os.path.join(root, filename)
except:
continue</code></pre>
</details>
</dd>
<dt id="ktrain.text.load_text_files"><code class="name flex">
<span>def <span class="ident">load_text_files</span></span>(<span>corpus_path, truncate_len=None, clean=True, return_fnames=False)</span>
</code></dt>
<dd>
<div class="desc"><pre><code>load text files
</code></pre></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def load_text_files(corpus_path, truncate_len=None,
clean=True, return_fnames=False):
"""
```
load text files
```
"""
texts = []
filenames = []
mb = master_bar(range(1))
for i in mb:
for filename in progress_bar(list(extract_filenames(corpus_path)), parent=mb):
with open(filename, 'r') as f:
text = f.read()
if clean:
text = strip_control_characters(text)
text = to_ascii(text)
if truncate_len is not None:
text = " ".join(text.split()[:truncate_len])
texts.append(text)
filenames.append(filename)
mb.write('done.')
if return_fnames:
return (texts, filenames)
else:
return texts</code></pre>
</details>
</dd>
<dt id="ktrain.text.print_sequence_taggers"><code class="name flex">
<span>def <span class="ident">print_sequence_taggers</span></span>(<span>)</span>
</code></dt>
<dd>
<div class="desc"></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def print_sequence_taggers():
for k,v in SEQUENCE_TAGGERS.items():
print("%s: %s" % (k,v))</code></pre>
</details>
</dd>
<dt id="ktrain.text.print_text_classifiers"><code class="name flex">
<span>def <span class="ident">print_text_classifiers</span></span>(<span>)</span>
</code></dt>
<dd>
<div class="desc"></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def print_text_classifiers():
for k,v in TEXT_CLASSIFIERS.items():
print("%s: %s" % (k,v))</code></pre>
</details>
</dd>
<dt id="ktrain.text.print_text_regression_models"><code class="name flex">
<span>def <span class="ident">print_text_regression_models</span></span>(<span>)</span>
</code></dt>
<dd>
<div class="desc"></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def print_text_regression_models():
for k,v in TEXT_REGRESSION_MODELS.items():
print("%s: %s" % (k,v))</code></pre>
</details>
</dd>
<dt id="ktrain.text.sequence_tagger"><code class="name flex">
<span>def <span class="ident">sequence_tagger</span></span>(<span>name, preproc, wv_path_or_url=None, bert_model='bert-base-multilingual-cased', bert_layers_to_use=[-2], word_embedding_dim=100, char_embedding_dim=25, word_lstm_size=100, char_lstm_size=25, fc_dim=100, dropout=0.5, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><p>Build and return a sequence tagger (i.e., named entity recognizer).</p>
<h2 id="args">Args</h2>
<dl>
<dt><strong><code>name</code></strong> : <code>string</code></dt>
<dd>one of:
- 'bilstm-crf' for Bidirectional LSTM-CRF model
- 'bilstm' for Bidirectional LSTM (no CRF layer)</dd>
</dl>
<p>preproc(NERPreprocessor):
an instance of NERPreprocessor
wv_path_or_url(str): either a URL or file path toa fasttext word vector file (.vec or .vec.zip or .vec.gz)
Example valid values for wv_path_or_url:</p>
<pre><code> Randomly-initialized word embeeddings:
set wv_path_or_url=None
English pretrained word vectors:
<https://dl.fbaipublicfiles.com/fasttext/vectors-english/crawl-300d-2M.vec.zip>
Chinese pretrained word vectors:
<https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.zh.300.vec.gz>
Russian pretrained word vectors:
<https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.ru.300.vec.gz>
Dutch pretrained word vectors:
<https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.nl.300.vec.gz>
See these two Web pages for a full list of URLs to word vector files for
different languages:
1. <https://fasttext.cc/docs/en/english-vectors.html> (for English)
2. <https://fasttext.cc/docs/en/crawl-vectors.html> (for non-English langages)
Default:None (randomly-initialized word embeddings are used)
</code></pre>
<p>bert_model_name(str):
the name of the BERT model.
default: 'bert-base-multilingual-cased'
This parameter is only used if bilstm-bert is selected for name parameter.
The value of this parameter is a name of BERT model from here:
<a href="https://huggingface.co/transformers/pretrained_models.html">https://huggingface.co/transformers/pretrained_models.html</a>
or a community-uploaded BERT model from here:
<a href="https://huggingface.co/models">https://huggingface.co/models</a>
Example values:
bert-base-multilingual-cased:
Multilingual BERT (157 languages) - this is the default
bert-base-cased:
English BERT
bert-base-chinese: Chinese BERT
distilbert-base-german-cased: German DistilBert
albert-base-v2: English ALBERT model
monologg/biobert_v1.1_pubmed: community uploaded BioBERT (pretrained on PubMed)</p>
<dl>
<dt>bert_layers_to_use(list): indices of hidden layers to use.
default:[-2] # second-to-last layer</dt>
<dt>To use the concatenation of last 4 layers: use [-1, -2, -3, -4]</dt>
<dt><strong><code>word_embedding_dim</code></strong> : <code>int</code></dt>
<dd>word embedding dimensions.</dd>
<dt><strong><code>char_embedding_dim</code></strong> : <code>int</code></dt>
<dd>character embedding dimensions.</dd>
<dt><strong><code>word_lstm_size</code></strong> : <code>int</code></dt>
<dd>character LSTM feature extractor output dimensions.</dd>
<dt><strong><code>char_lstm_size</code></strong> : <code>int</code></dt>
<dd>word tagger LSTM output dimensions.</dd>
<dt><strong><code>fc_dim</code></strong> : <code>int</code></dt>
<dd>output fully-connected layer size.</dd>
<dt><strong><code>dropout</code></strong> : <code>float</code></dt>
<dd>dropout rate.</dd>
<dt><strong><code>verbose</code></strong> : <code>boolean</code></dt>
<dd>verbosity of output</dd>
</dl>
<h2 id="return">Return</h2>
<p>model (Model): A Keras Model instance</p></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def sequence_tagger(name, preproc,
wv_path_or_url=None,
bert_model = 'bert-base-multilingual-cased',
bert_layers_to_use = U.DEFAULT_TRANSFORMER_LAYERS,
word_embedding_dim=100,
char_embedding_dim=25,
word_lstm_size=100,
char_lstm_size=25,
fc_dim=100,
dropout=0.5,
verbose=1):
"""
Build and return a sequence tagger (i.e., named entity recognizer).
Args:
name (string): one of:
- 'bilstm-crf' for Bidirectional LSTM-CRF model
- 'bilstm' for Bidirectional LSTM (no CRF layer)
preproc(NERPreprocessor): an instance of NERPreprocessor
wv_path_or_url(str): either a URL or file path toa fasttext word vector file (.vec or .vec.zip or .vec.gz)
Example valid values for wv_path_or_url:
Randomly-initialized word embeeddings:
set wv_path_or_url=None
English pretrained word vectors:
https://dl.fbaipublicfiles.com/fasttext/vectors-english/crawl-300d-2M.vec.zip
Chinese pretrained word vectors:
https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.zh.300.vec.gz
Russian pretrained word vectors:
https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.ru.300.vec.gz
Dutch pretrained word vectors:
https://dl.fbaipublicfiles.com/fasttext/vectors-crawl/cc.nl.300.vec.gz
See these two Web pages for a full list of URLs to word vector files for
different languages:
1. https://fasttext.cc/docs/en/english-vectors.html (for English)
2. https://fasttext.cc/docs/en/crawl-vectors.html (for non-English langages)
Default:None (randomly-initialized word embeddings are used)
bert_model_name(str): the name of the BERT model. default: 'bert-base-multilingual-cased'
This parameter is only used if bilstm-bert is selected for name parameter.
The value of this parameter is a name of BERT model from here:
https://huggingface.co/transformers/pretrained_models.html
or a community-uploaded BERT model from here:
https://huggingface.co/models
Example values:
bert-base-multilingual-cased: Multilingual BERT (157 languages) - this is the default
bert-base-cased: English BERT
bert-base-chinese: Chinese BERT
distilbert-base-german-cased: German DistilBert
albert-base-v2: English ALBERT model
monologg/biobert_v1.1_pubmed: community uploaded BioBERT (pretrained on PubMed)
bert_layers_to_use(list): indices of hidden layers to use. default:[-2] # second-to-last layer
To use the concatenation of last 4 layers: use [-1, -2, -3, -4]
word_embedding_dim (int): word embedding dimensions.
char_embedding_dim (int): character embedding dimensions.
word_lstm_size (int): character LSTM feature extractor output dimensions.
char_lstm_size (int): word tagger LSTM output dimensions.
fc_dim (int): output fully-connected layer size.
dropout (float): dropout rate.
verbose (boolean): verbosity of output
Return:
model (Model): A Keras Model instance
"""
if name not in SEQUENCE_TAGGERS:
raise ValueError('invalid name: %s' % (name))
# check BERT
if name in TRANSFORMER_MODELS and not bert_model:
raise ValueError('bert_model is required for bilstm-bert models')
if name in TRANSFORMER_MODELS and DISABLE_V2_BEHAVIOR:
raise ValueError('BERT and other transformer models cannot be used with DISABLE_v2_BEHAVIOR')
# check CRF
if not DISABLE_V2_BEHAVIOR and name in V1_ONLY_MODELS:
warnings.warn('Falling back to BiLSTM (no CRF) because DISABLE_V2_BEHAVIOR=False')
msg = "\nIMPORTANT NOTE: ktrain uses the CRF module from keras_contrib, which is not yet\n" +\
"fully compatible with TensorFlow 2. You can still use the BiLSTM-CRF model\n" +\
"in ktrain for sequence tagging with TensorFlow 2, but you must add the\n" +\
"following to the top of your script or notebook BEFORE you import ktrain:\n\n" +\
"import os\n" +\
"os.environ['DISABLE_V2_BEHAVIOR'] = '1'\n\n" +\
"For this run, a vanilla BiLSTM model (with no CRF layer) will be used.\n"
print(msg)
name = BILSTM if name == BILSTM_CRF else BILSTM_ELMO
# check for use_char=True
if not DISABLE_V2_BEHAVIOR and preproc.p._use_char:
# turn off masking due to open TF2 issue ##33148: https://github.com/tensorflow/tensorflow/issues/33148
warnings.warn('Setting use_char=False: character embeddings cannot be used in TF2 due to open TensorFlow 2 bug (#33148).\n' +\
'Add os.environ["DISABLE_V2_BEHAVIOR"] = "1" to the top of script if you really want to use it.')
preproc.p._use_char=False
if verbose:
emb_names = []
if wv_path_or_url is not None:
emb_names.append('word embeddings initialized with fasttext word vectors (%s)' % (os.path.basename(wv_path_or_url)))
else:
emb_names.append('word embeddings initialized randomly')
if name in TRANSFORMER_MODELS: emb_names.append('BERT embeddings with ' + bert_model)
if name in ELMO_MODELS: emb_names.append('Elmo embeddings for English')
if preproc.p._use_char: emb_names.append('character embeddings')
if len(emb_names) > 1:
print('Embedding schemes employed (combined with concatenation):')
else:
print('embedding schemes employed:')
for emb_name in emb_names:
print('\t%s' % (emb_name))
print()
# setup embedding
if wv_path_or_url is not None:
wv_model, word_embedding_dim = preproc.get_wv_model(wv_path_or_url, verbose=verbose)
else:
wv_model = None
if name == BILSTM_CRF:
use_crf = False if not DISABLE_V2_BEHAVIOR else True # fallback to bilstm
elif name == BILSTM_CRF_ELMO:
use_crf = False if not DISABLE_V2_BEHAVIOR else True # fallback to bilstm
preproc.p.activate_elmo()
elif name == BILSTM:
use_crf = False
elif name == BILSTM_ELMO:
use_crf = False
preproc.p.activate_elmo()
elif name == BILSTM_TRANSFORMER:
use_crf = False
preproc.p.activate_transformer(bert_model, layers=bert_layers_to_use, force=True)
else:
raise ValueError('Unsupported model name')
model = BiLSTMCRF(char_embedding_dim=char_embedding_dim,
word_embedding_dim=word_embedding_dim,
char_lstm_size=char_lstm_size,
word_lstm_size=word_lstm_size,
fc_dim=fc_dim,
char_vocab_size=preproc.p.char_vocab_size,
word_vocab_size=preproc.p.word_vocab_size,
num_labels=preproc.p.label_size,
dropout=dropout,
use_crf=use_crf,
use_char=preproc.p._use_char,
embeddings=wv_model,
use_elmo=preproc.p.elmo_is_activated(),
use_transformer_with_dim=preproc.p.get_transformer_dim())
model, loss = model.build()
model.compile(loss=loss, optimizer=U.DEFAULT_OPT)
return model</code></pre>
</details>
</dd>
<dt id="ktrain.text.text_classifier"><code class="name flex">
<span>def <span class="ident">text_classifier</span></span>(<span>name, train_data, preproc=None, multilabel=None, metrics=['accuracy'], verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><pre><code>Build and return a text classification model.
Args:
name (string): one of:
- 'fasttext' for FastText model
- 'nbsvm' for NBSVM model
- 'logreg' for logistic regression using embedding layers
- 'bigru' for Bidirectional GRU with pretrained word vectors
- 'bert' for BERT Text Classification
- 'distilbert' for Hugging Face DistilBert model
train_data (tuple): a tuple of numpy.ndarrays: (x_train, y_train) or ktrain.Dataset instance
returned from one of the texts_from_* functions
preproc: a ktrain.text.TextPreprocessor instance.
As of v0.8.0, this is required.
multilabel (bool): If True, multilabel model will be returned.
If false, binary/multiclass model will be returned.
If None, multilabel will be inferred from data.
metrics(list): metrics to use
verbose (boolean): verbosity of output
Return:
model (Model): A Keras Model instance
</code></pre></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def text_classifier(name, train_data, preproc=None, multilabel=None, metrics=['accuracy'], verbose=1):
"""
```
Build and return a text classification model.
Args:
name (string): one of:
- 'fasttext' for FastText model
- 'nbsvm' for NBSVM model
- 'logreg' for logistic regression using embedding layers
- 'bigru' for Bidirectional GRU with pretrained word vectors
- 'bert' for BERT Text Classification
- 'distilbert' for Hugging Face DistilBert model
train_data (tuple): a tuple of numpy.ndarrays: (x_train, y_train) or ktrain.Dataset instance
returned from one of the texts_from_* functions
preproc: a ktrain.text.TextPreprocessor instance.
As of v0.8.0, this is required.
multilabel (bool): If True, multilabel model will be returned.
If false, binary/multiclass model will be returned.
If None, multilabel will be inferred from data.
metrics(list): metrics to use
verbose (boolean): verbosity of output
Return:
model (Model): A Keras Model instance
```
"""
if name not in TEXT_CLASSIFIERS:
raise ValueError('invalid name for text classification: %s' % (name))
if preproc is not None and not preproc.get_classes():
raise ValueError('preproc.get_classes() is empty, but required for text classification')
return _text_model(name, train_data, preproc=preproc,
multilabel=multilabel, classification=True, metrics=metrics, verbose=verbose)</code></pre>
</details>
</dd>
<dt id="ktrain.text.text_regression_model"><code class="name flex">
<span>def <span class="ident">text_regression_model</span></span>(<span>name, train_data, preproc=None, metrics=['mae'], verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><pre><code>Build and return a text regression model.
Args:
name (string): one of:
- 'fasttext' for FastText model
- 'nbsvm' for NBSVM model
- 'linreg' for linear regression using embedding layers
- 'bigru' for Bidirectional GRU with pretrained word vectors
- 'bert' for BERT Text Classification
- 'distilbert' for Hugging Face DistilBert model
train_data (tuple): a tuple of numpy.ndarrays: (x_train, y_train)
preproc: a ktrain.text.TextPreprocessor instance.
As of v0.8.0, this is required.
metrics(list): metrics to use
verbose (boolean): verbosity of output
Return:
model (Model): A Keras Model instance
</code></pre></div>
<details class="source">
<summary>
<span>Expand source code</span>
</summary>
<pre><code class="python">def text_regression_model(name, train_data, preproc=None, metrics=['mae'], verbose=1):
"""
```
Build and return a text regression model.
Args:
name (string): one of:
- 'fasttext' for FastText model
- 'nbsvm' for NBSVM model
- 'linreg' for linear regression using embedding layers
- 'bigru' for Bidirectional GRU with pretrained word vectors
- 'bert' for BERT Text Classification
- 'distilbert' for Hugging Face DistilBert model
train_data (tuple): a tuple of numpy.ndarrays: (x_train, y_train)
preproc: a ktrain.text.TextPreprocessor instance.
As of v0.8.0, this is required.
metrics(list): metrics to use
verbose (boolean): verbosity of output
Return:
model (Model): A Keras Model instance
```
"""
if name not in TEXT_REGRESSION_MODELS:
raise ValueError('invalid name for text classification: %s' % (name) )
if preproc is not None and preproc.get_classes():
raise ValueError('preproc.get_classes() is supposed to be empty for text regression tasks')
return _text_model(name, train_data, preproc=preproc,
multilabel=False, classification=False, metrics=metrics, verbose=verbose)</code></pre>
</details>
</dd>
<dt id="ktrain.text.texts_from_array"><code class="name flex">
<span>def <span class="ident">texts_from_array</span></span>(<span>x_train, y_train, x_test=None, y_test=None, class_names=[], max_features=20000, maxlen=400, val_pct=0.1, ngram_range=1, preprocess_mode='standard', lang=None, random_state=None, verbose=1)</span>
</code></dt>
<dd>
<div class="desc"><pre><code>Loads and preprocesses text data from arrays.
texts_from_array can handle data for both text classification
and text regression. If class_names is empty, a regression task is assumed.
Args:
x_train(list): list of training texts
y_train(list): labels in one of the following forms:
1. list of integers representing classes (class_names is required)