-
Notifications
You must be signed in to change notification settings - Fork 0
/
PlotGenerator.py
1169 lines (926 loc) · 42 KB
/
PlotGenerator.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
# -*- coding: utf-8 -*-
"""21L-6277_21L-5631.ipynb
Automatically generated by Colab.
Original file is located at
https://colab.research.google.com/drive/17GaNsDYMNV-O3Q1ekaqOxHJsWuFHrX_P
# Genre Pre-Processing
"""
import re
import pandas as pd
import numpy as np
import pandas as pd
import nltk
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
from google.colab import drive
drive.mount('/content/drive')
path = "/content/drive/My Drive/movieplots.csv"
df = pd.read_csv(path)
df
df = pd.read_csv('movieplots.csv')
df
print(df['Plots'][14])
genrecounts = df['Genre'].value_counts()
print(genrecounts)
df['Genre'] = df['Genre'].str.strip()
df['Genre'].unique()
# Drop rows with genre 'unknown'
df = df[df['Genre'] != 'unknown']
count = df['Genre'].value_counts()
print(count)
df.shape
print(df['Genre'].unique())
def map_genre(genre):
if isinstance(genre, str): # Check if genre is a string
genre = genre.lower()
if any(substring in genre for substring in ['drama', 'melodrama', 'docudrama', 'romantic drama', 'historical drama', 'comedy-drama', 'crime drama', 'war drama', 'action drama', 'sports drama', 'family drama', 'teen drama', 'political drama', 'legal drama', 'medical drama', 'anthology drama']):
return 'drama'
elif any(substring in genre for substring in ['action', 'action adventure', 'action comedy', 'action thriller', 'action drama', 'action horror', 'action mystery', 'action romance', 'action sci-fi', 'action spy', 'action war', 'action western']):
return 'action'
elif any(substring in genre for substring in ['romance', 'romantic comedy', 'romantic drama', 'romantic fantasy', 'romantic thriller']):
return 'romance'
elif any(substring in genre for substring in ['comedy', 'romantic comedy', 'comedy drama', 'comedy horror', 'comedy mystery', 'comedy thriller', 'action comedy', 'black comedy', 'sex comedy', 'slapstick comedy', 'stand-up comedy']):
return 'comedy'
elif any(substring in genre for substring in ['horror', 'horror comedy', 'horror drama', 'horror thriller', 'action horror', 'comedy horror', 'psychological horror', 'supernatural horror']):
return 'horror'
elif any(substring in genre for substring in ['thriller', 'action thriller', 'comedy thriller', 'crime thriller', 'horror thriller', 'psychological thriller', 'supernatural thriller']):
return 'thriller'
elif any(substring in genre for substring in ['anime', 'anime drama', 'anime fantasy', 'anime horror', 'anime mystery', 'anime romance', 'anime sci-fi']):
return 'anime'
elif any(substring in genre for substring in ['mystery', 'action mystery', 'comedy mystery', 'crime mystery', 'horror mystery', 'mystery thriller']):
return 'mystery'
elif any(substring in genre for substring in ['sci-fi', 'action sci-fi', 'anime sci-fi', 'comedy sci-fi', 'drama sci-fi', 'horror sci-fi', 'mystery sci-fi', 'romance sci-fi', 'sci-fi thriller']):
return 'sci-fi'
return genre # Return the original genre if it's not a string or NaN
df['Genre'] = df['Genre'].apply(map_genre)
print(df['Genre'].nunique())
def map_genre(genre):
if isinstance(genre, str): # Check if genre is a string
genre = genre.lower()
if 'sci-fi' in genre or 'science fiction' in genre or 'sci-fi for children' in genre:
return 'Sci-Fi'
elif 'romantic comedy' in genre or 'rom-com' in genre or 'comedy / drama / romance' in genre or 'romantic comedy-drama' in genre or 'romance/comedy' in genre or 'comedy/romance' in genre:
return 'Romantic Comedy'
elif 'biopic' in genre or 'biography' in genre or 'biogtaphy' in genre or 'bio-pic, sports' in genre or 'bio-pic, drama, music' in genre or 'biographical, documentary' in genre or 'biographical, drama' in genre or 'biography / drama / romance' in genre or 'biography crime' in genre or 'biographical war film' in genre:
return 'Biography'
elif 'animation' in genre or 'anime' in genre or 'cartoon' in genre or 'animation adventure' in genre or 'animation comedy' in genre or 'animation fantasy' in genre or 'animated adventure' in genre or 'animation, adventure, sci-fi' in genre or 'animation, comedy, action' in genre or 'animation, live action / drama / comedy' in genre or 'animation / adventure' in genre or 'animation martial arts action-comedy' in genre or 'animation musical' in genre:
return 'Animation'
elif 'horror' in genre or 'horror-thrill' in genre or 'horror-thriller' in genre or 'horror / mystery / thriller' in genre or 'horror, thriller, drama' in genre or 'horror, sci-fi, drama' in genre or 'horror comedy, parody' in genre:
return 'Horror'
else:
return genre # return the original genre if no match is found
df['Genre'] = df['Genre'].apply(map_genre)
print(df['Genre'].nunique())
print(df['Genre'].unique())
def map_genre(genre):
if isinstance(genre, str): # Check if genre is a string
genre = genre.lower()
if 'horror' in genre:
return 'Horror'
elif 'thriller' in genre or 'suspense' in genre or 'neo-noir' in genre or 'spy' in genre or 'espionage' in genre or 'spy film' in genre or 'spy serial' in genre or 'spy anthology' in genre or 'spy spoof' in genre:
return 'Thriller'
elif 'scifi' in genre or 'science-fiction' in genre or 'Sci-Fi' in genre:
return 'Sci-Fi'
elif 'action' in genre or 'adventures' in genre or 'outlaw biker film' in genre or 'biker film' in genre or 'disaster film' in genre or 'martial arts' in genre or 'disaster' in genre:
return 'Action'
elif 'romance' in genre or 'romantic' in genre or 'romantic musical' in genre:
return 'Romance'
elif 'western' in genre or 'western, musical' in genre or 'western, war' in genre or 'western musical' in genre or 'western serial' in genre or 'western, serial' in genre or 'western, film noir' in genre or 'western, 3-d' in genre or 'musical, western' in genre or 'war, western' in genre:
return 'Western'
elif 'comedy' in genre or 'rom com' in genre or 'spy spoof' in genre or 'mockumentary' in genre:
return 'Comedy'
elif 'short' in genre or 'short film' in genre or 'short fantasy' in genre or 'short subject' in genre:
return 'Short'
elif 'biographical' in genre or 'biography' in genre or 'bio-pic' in genre or 'musical bio-pic' in genre:
return 'Biography'
elif 'drama' in genre:
return 'Drama'
elif 'adventure' in genre or 'adventures' in genre or 'adventure serial' in genre or 'adventure film' in genre or 'adventure, fantasy' in genre or 'adventure, music' in genre or 'adventure, serial' in genre:
return 'Adventure'
elif 'fantasy' in genre or 'short fantasy' in genre or 'fantasy, family' in genre or 'fantasy, adventure' in genre or 'musical fantasy' in genre or 'animated, fantasy':
return 'Fantasy'
elif 'crime' in genre or 'true crime' in genre or 'crime musical' in genre or 'crime, adventure' in genre or 'crime, film noir' in genre or 'crime, western' in genre or 'gangster':
return 'Crime'
elif 'documentary' in genre or 'semi-staged documentary' in genre or 'propaganda short' in genre or 'propaganda short animated short' in genre or 'compilation' in genre or 'experimental short' in genre or 'propaganda' in genre or 'ww1 propaganda' in genre or 'war documentary' in genre or 'sexual hygiene/exploitation film' in genre or 'exploitation' in genre or 'sexploitation' in genre or 'adult' in genre or 'adult film' in genre or 'erotic musical':
return 'Documentary'
elif 'family' in genre or 'fantasy, family' in genre or 'family, western' in genre or 'musical, family' in genre or 'animated, family':
return 'Family'
elif 'musical' in genre or 'operetta' in genre or 'musical western' in genre or 'animated, musical' in genre or 'musical fantasy' in genre or 'war short' in genre or 'animated musical' in genre or 'musical, fantasy' in genre or 'musical, family' in genre or 'musical, western' in genre or 'erotic musical':
return 'Musical'
elif 'war' in genre or 'war spy' in genre or 'war documentary' in genre or 'war propaganda' in genre or 'war, satire' in genre or 'war, western' in genre or 'war, biker':
return 'War'
elif 'epic' in genre or 'historic' in genre or 'historical' in genre or 'historical epic' in genre or 'paramount. biblical epic':
return 'History'
elif 'silent sports' in genre or 'sports' in genre or 'american football':
return 'Sport'
elif 'animation' in genre or 'animated series' in genre or 'Animation' in genre or 'animated' in genre or 'animated, musical' in genre or 'animated short' in genre or 'animated film' in genre or 'animated, war' in genre or 'animated, adult':
return 'Animation'
elif 'mystery' in genre or 'detective' in genre or 'charlie chan':
return 'Mystery'
else:
return genre
df['Genre'] = df['Genre'].apply(map_genre)
print(df['Genre'].nunique())
print(df['Genre'].unique())
df.shape
print(df['Genre'].value_counts())
"""# Plots Preprocessing
**Droping null values**
"""
print(df.isnull().sum())
df = df.dropna(subset=['Plots'])
df = df.dropna(subset=['Genre'])
print(df.isnull().sum())
"""**REMOVING SPECIAL CHARACTERS**"""
def remove_special_characters(text):
text = re.sub(r'[^a-zA-Z0-9\s]', '', text)
return text
df['Plots'] = df['Plots'].apply(remove_special_characters)
"""**CONVERSION TO LOWER CASE**"""
def convert_to_lowercase(text):
return text.lower()
df['Plots'] = df['Plots'].apply(convert_to_lowercase)
df.shape
nltk.download('stopwords')
nltk.download('punkt')
"""**REMOVING STOP WORDS (AS AN EXPERIMENT AS IN I MIGHT OR MIGHT NOT RUN THIS DEPENDING ON THE PERPOMANCE OF THE CODE)**"""
from nltk.corpus import stopwords
from nltk.tokenize import word_tokenize
stop_words = set(stopwords.words('english'))
def remove_stopwords(text):
tokens = word_tokenize(text)
filtered_tokens = [word for word in tokens if word.lower() not in stop_words]
return ' '.join(filtered_tokens)
df['Plots'] = df['Plots'].apply(remove_stopwords)
df
"""**TOKENIZATION**"""
import spacy
nlp = spacy.load('en_core_web_sm', disable = ['parser', 'tagger', 'ner'])
def get_tokens(doc_text):
skip_pattern = '\r\n \n\n \n\n\n!"-#$%&()--.*+,-./:;<=>?@[\\]^_`{|}~\t\n\r '
tokens = [token.text.lower() for token in nlp(doc_text) if token.text not in skip_pattern]
return tokens
df['tokens'] = df['Plots'].apply(get_tokens)
df
"""**Sequence Creation:**"""
#this is explicitly present in model now so dont run rn
train_len = 25 + 1 # 25 words input, 1 word output
text_sequences = df['Plots'].tolist()
for i in range(train_len, len(df['tokens'])):
seq = df['tokens'][i - train_len: i]
text_sequences.append(seq)
"""**Vectorization**"""
#this is explicitly present in model now so dont run rn
# Convert the 'tokens' column to a list of lists
tokens = df['tokens'].tolist()
# Define the length of your sequences
train_len = 25 + 1 # 25 words input, 1 word output
text_sequences = []
# Create the sequences
for i in range(train_len, len(tokens)):
seq = tokens[i - train_len: i]
text_sequences.append(seq)
# Flatten the list of tokens and convert it into a string
text_sequences = [' '.join(token) for sublist in text_sequences for token in sublist]
# Now you can fit the tokenizer on the text sequences
tokenizer = Tokenizer()
tokenizer.fit_on_texts(text_sequences)
sequences = tokenizer.texts_to_sequences(text_sequences)
"""**RARE WORDS** **(THIS TOO WILL BE RUN ON TRIAL BASIS)**"""
from collections import Counter
all_words = ' '.join(df['Plots']).split()
word_counts = Counter(all_words)
threshold = 5 # Adjust this value based on your dataset and requirements
rare_words = [word for word, count in word_counts.items() if count <= threshold]
def handle_rare_words(text, rare_words, replacement='<UNK>'):
tokens = text.split()
filtered_tokens = [token if token not in rare_words else replacement for token in tokens]
return ' '.join(filtered_tokens)
df['Plots'] = df['Plots'].apply(lambda x: handle_rare_words(x, rare_words))
df
path = "/content/drive/My Drive/Plot.csv"
df = pd.read_csv(path)
df
df.to_csv('Plot.csv', index=False)
"""# **VISUALISATION**"""
import matplotlib.pyplot as plt
plt.figure(figsize=(10, 6))
df['Genre'].value_counts().plot(kind='bar')
plt.title('Genre Distribution')
plt.xlabel('Genre')
plt.ylabel('Count')
plt.show()
plt.figure(figsize=(10, 6))
df['Plots'].apply(lambda x: len(x.split())).plot(kind='hist', bins=30)
plt.title('Distribution of Plot Lengths')
plt.xlabel('Number of Words')
plt.ylabel('Frequency')
plt.show()
from wordcloud import WordCloud
wordcloud = WordCloud(width=800, height=400, background_color='white').generate(' '.join(df['Plots']))
plt.figure(figsize=(12, 6))
plt.imshow(wordcloud, interpolation='bilinear')
plt.title('Word Cloud of Plot Descriptions')
plt.axis('off')
plt.show()
import seaborn as sns
plt.figure(figsize=(12, 6))
df['word_count'] = df['Plots'].apply(lambda x: len(x.split()))
sns.boxplot(x='Genre', y='word_count', data=df)
plt.title('Distribution of Plot Lengths by Genre')
plt.xlabel('Genre')
plt.ylabel('Number of Words')
plt.xticks(rotation=45)
plt.show()
import seaborn as sns
import matplotlib.pyplot as plt
binary_matrix = pd.get_dummies(df['Genre'])
correlation_matrix = binary_matrix.corr()
plt.figure(figsize=(10, 8))
sns.heatmap(correlation_matrix, annot=True, cmap='coolwarm', fmt=".2f", square=True)
plt.title('Genre Correlation Heatmap')
plt.xlabel('Genre')
plt.ylabel('Genre')
plt.show()
df
"""# **MODEL FITTING**
# RNN
"""
import spacy
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
from keras.models import Sequential
from keras.layers import SimpleRNN, Embedding, Dense
from keras.utils import to_categorical
# Convert the 'tokens' column to a list of lists
tokens = df['tokens'].tolist()[:1000] # Use only the first 1,000 plots
# Define the length of your sequences
train_len = 25 + 1 # 25 words input, 1 word output
text_sequences = []
# Create the sequences
for i in range(train_len, len(tokens)):
seq = tokens[i - train_len: i]
text_sequences.append(seq)
# Flatten the list of tokens and convert it into a string
text_sequences = [' '.join(token) for sublist in text_sequences for token in sublist]
# Now you can fit the tokenizer on the text sequences
tokenizer = Tokenizer()
tokenizer.fit_on_texts(text_sequences)
sequences = tokenizer.texts_to_sequences(text_sequences)
# Vocabulary size
vocabulary_size = len(tokenizer.word_index) + 1
# Padding
sequences = pad_sequences(sequences, maxlen=train_len, truncating='pre')
# Split sequences into input (X) and output (y)
X = sequences[:, :-1]
y = sequences[:, -1]
y = to_categorical(y, num_classes=vocabulary_size)
from keras.layers import Dropout
model = Sequential()
model.add(Embedding(input_dim=vocabulary_size, output_dim=100, input_length=train_len-1))
model.add(SimpleRNN(256, return_sequences=True))
model.add(Dropout(0.2))
model.add(SimpleRNN(128, return_sequences=True))
model.add(Dropout(0.2))
model.add(SimpleRNN(64))
model.add(Dropout(0.2))
model.add(Dense(vocabulary_size, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
print(model.summary())
# Train the model
model.fit(X, y, batch_size=128, epochs=10, verbose=1)
def generate_text(key_words, next_words, max_sequence_len):
seed_text = ' '.join(key_words)
generated_text = seed_text
for _ in range(next_words):
token_list = tokenizer.texts_to_sequences([seed_text])[0]
token_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')
predicted_probs = model.predict(token_list, verbose=0)[0]
predicted_index = np.argmax(predicted_probs) + 1
output_word = ""
for word, index in tokenizer.word_index.items():
if index == predicted_index:
output_word = word
break
seed_text += " " + output_word
generated_text += " " + output_word
return generated_text
new_plot = ["mystery", "haunted", "dark", "fear"]
generated_text = generate_text(new_plot, 20, train_len)
print(generated_text)
import numpy as np
# Function to generate text based on genre
def generate_plot(genre, next_words, max_sequence_len):
# Include genre in the seed text
seed_text = genre.lower()
generated_text = seed_text
for _ in range(next_words):
token_list = tokenizer.texts_to_sequences([seed_text])[0]
token_list = pad_sequences([token_list], maxlen=max_sequence_len-1, padding='pre')
predicted_probs = model.predict(token_list, verbose=0)[0]
predicted_index = np.argmax(predicted_probs) + 1
output_word = ""
for word, index in tokenizer.word_index.items():
if index == predicted_index:
output_word = word
break
seed_text += " " + output_word
generated_text += " " + output_word
return generated_text
# Example usage
genre = "drama"
generated_plot = generate_plot(genre, 20, train_len)
print(generated_plot)
# Split the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Evaluate the model on the testing set
loss, accuracy = model.evaluate(X_test, y_test, verbose=1)
print("Test Accuracy:", accuracy)
"""# LSTM(WITH SOME CUSTOMISATION)"""
from keras.models import Sequential
from keras.layers import Dense, LSTM, Embedding
from keras.utils import to_categorical
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
tokens = df['tokens'].tolist()[:10000] # Use only the first 1,000 plots
train_len = 25 + 1
text_sequences = []
for i in range(train_len, len(tokens)):
seq = tokens[i - train_len: i]
text_sequences.append(seq)
text_sequences = [' '.join(map(str, seq)) for seq in text_sequences]
tokenizer = Tokenizer()
tokenizer.fit_on_texts(text_sequences)
sequences = tokenizer.texts_to_sequences(text_sequences)
vocabulary_size = len(tokenizer.word_index) + 1
sequences = pad_sequences(sequences, maxlen=train_len, truncating='pre')
X = sequences[:, :-1]
y = sequences[:, -1]
y = to_categorical(y, num_classes=vocabulary_size)
# Model creation
model = Sequential()
model.add(Embedding(input_dim=vocabulary_size, output_dim=train_len-1, input_length=train_len-1))
model.add(LSTM(units=50, return_sequences=True))
model.add(LSTM(units=50))
model.add(Dense(units=50, activation='relu'))
model.add(Dense(units=vocabulary_size, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
# Training
model.fit(X, y, batch_size=128, epochs=80)
import numpy as np
def generate_plot(model, tokenizer, seq_len, seed_text, num_gen_words):
output_text = []
input_text = seed_text
for _ in range(num_gen_words):
encoded_text = tokenizer.texts_to_sequences([input_text])[0]
# padding matching input length check
pad_encoded = pad_sequences([encoded_text], maxlen=seq_len-1, truncating='pre')
pred_word_ind = np.argmax(model.predict(pad_encoded), axis=-1)[0]
pred_word = tokenizer.index_word.get(pred_word_ind, '')
input_text += ' ' + pred_word
output_text.append(pred_word)
return ' '.join(output_text)
start = "a dark mysterious figure lurking through the"
no = 100
seq_len = 25
newplot = generate_plot(model, tokenizer, seq_len,start,no)
print(newplot)
"""# CUSTOM EMBEDINGS"""
import numpy as np
from keras.models import Sequential
from keras.layers import LSTM, Dense, Embedding, Dropout
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
from keras.utils import to_categorical
from sklearn.model_selection import train_test_split
seq_len = 50
tokenizer = Tokenizer()
tokenizer.fit_on_texts(df['tokens'].tolist()[:10000])
sequences = tokenizer.texts_to_sequences(df['tokens'].tolist()[:10000])
# Padding
sequences = pad_sequences(sequences, maxlen=seq_len, padding='pre')
# Split sequences into input (X) and output (y)
X, y = sequences[:, :-1], sequences[:, -1]
y = to_categorical(y, num_classes=len(tokenizer.word_index) + 1)
# Spliting the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=42)
def create_lstm_model(seq_len, vocabulary_size):
model = Sequential()
model.add(Embedding(input_dim=vocabulary_size, output_dim=100, input_length=seq_len-1))
model.add(LSTM(units=128, return_sequences=True))
model.add(Dropout(0.2))
model.add(LSTM(units=128))
model.add(Dropout(0.2))
model.add(Dense(units=vocabulary_size, activation='softmax'))
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
return model
vocabulary_size = len(tokenizer.word_index) + 1
model = create_lstm_model(seq_len, vocabulary_size)
model.fit(X_train, y_train, batch_size=128, epochs=70, validation_data=(X_test, y_test))
def generate_plot(model, tokenizer, seq_len, seed_text, num_gen_words):
output_text = seed_text
for _ in range(num_gen_words):
encoded_text = tokenizer.texts_to_sequences([output_text])[0]
pad_encoded = pad_sequences([encoded_text], maxlen=seq_len-1, truncating='pre')
pred_word_ind = np.argmax(model.predict(pad_encoded), axis=-1)[0]
pred_word = tokenizer.index_word.get(pred_word_ind, '')
if not pred_word:
break # Stop if the predicted word is empty or unknown
output_text += ' ' + pred_word
return output_text
# Generate a new plot
seed_text = "A dark mysterious figure lurking"
num_gen_words = 100
new_plot = generate_plot(model, tokenizer, seq_len, seed_text, num_gen_words)
print(new_plot)
print(df['tokens'][:10000])
import numpy as np
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
from keras.utils import to_categorical
from sklearn.model_selection import train_test_split
from keras.models import Sequential
from keras.layers import Embedding, LSTM, Dropout, Dense
seq_len = 50
tokenizer = Tokenizer()
tokenizer.fit_on_texts(df['tokens'][:10000])
sequences = tokenizer.texts_to_sequences(df['tokens'][:10000])
sequences = pad_sequences(sequences, maxlen=seq_len, padding='pre')
X = sequences[:, :-1] # All tokens except the last as input
y = sequences[:, -1] # The last token as the target output
# One-hot encoding the output
y = to_categorical(y, num_classes=len(tokenizer.word_index) + 1)
# Take a random subset of 5000 entries for both X and y
np.random.seed(42)
subset_indices = np.random.choice(len(X), 5000, replace=False)
X_subset = X[subset_indices]
y_subset = y[subset_indices]
X_train, X_test, y_train, y_test = train_test_split(X_subset, y_subset, test_size=0.1, random_state=42)
model = Sequential()
model.add(Embedding(input_dim=len(tokenizer.word_index) + 1, output_dim=100, input_length=seq_len-1))
model.add(LSTM(units=128, return_sequences=True))
model.add(Dropout(0.2))
model.add(LSTM(units=128))
model.add(Dropout(0.2))
model.add(Dense(units=len(tokenizer.word_index) + 1, activation='softmax'))
# Compile the model
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
# Train the model on the subset
model.fit(X_train, y_train, batch_size=128, epochs=100, validation_data=(X_test, y_test))
def generate_plot(model, tokenizer, seq_len, genre_to_id, genre):
seed_text_index = np.random.choice(len(X_train))
seed_text_sequence = X_train[seed_text_index]
seed_text = ' '.join([tokenizer.index_word.get(i, '') for i in seed_text_sequence if i != 0])
num_gen_words = 30
genre_input = to_categorical([genre_to_id[genre]], num_classes=len(genre_to_id))
output_text = seed_text
for _ in range(num_gen_words):
encoded_text = tokenizer.texts_to_sequences([output_text])[0]
pad_encoded = pad_sequences([encoded_text], maxlen=seq_len-1, truncating='pre')
pred_word_ind = np.argmax(model.predict(pad_encoded), axis=-1)[0]
pred_word = tokenizer.index_word.get(pred_word_ind, '')
if not pred_word:
break
output_text += ' ' + pred_word
return output_text
# Example usage
genre_to_id = {
'horror': 0,
'thriller': 1,
'sci-fi': 2,
'action': 3,
'romance': 4,
'fantasy': 5,
'western': 6,
'comedy': 7,
'short': 8,
'biography': 9,
'drama': 10,
'adventure': 11
}
genre = 'drama'
new_plot = generate_plot(model, tokenizer, seq_len, genre_to_id, genre)
print("Generated Plot:")
print(new_plot)
genre_to_id = {
'horror': 0,
'thriller': 1,
'sci-fi': 2,
'action': 3,
'romance': 4,
'fantasy': 5,
'western': 6,
'comedy': 7,
'short': 8,
'biography': 9,
'drama': 10,
'adventure': 11
}
import numpy as np
from keras.preprocessing.text import Tokenizer
from keras.preprocessing.sequence import pad_sequences
from keras.utils import to_categorical
from sklearn.model_selection import train_test_split
from keras.models import Sequential
from keras.layers import Embedding, LSTM, Dropout, Dense, Input, concatenate
from keras.models import Model
seq_len = 30
tokenizer = Tokenizer()
tokenizer.fit_on_texts(df['tokens'][:5000]) # Use a smaller subset
sequences = tokenizer.texts_to_sequences(df['tokens'][:5000])
sequences = pad_sequences(sequences, maxlen=seq_len, padding='pre')
X = sequences[:, :-1] # All tokens except the last as input
y = sequences[:, -1] # The last token as the target output
# One-hot encoding the output
y = to_categorical(y, num_classes=len(tokenizer.word_index) + 1)
np.random.seed(42)
subset_indices = np.random.choice(len(X), 5000, replace=False) # Use a smaller subset
X_subset = X[subset_indices]
y_subset = y[subset_indices]
# Create genre labels (assuming 'df' has a 'Genre' column)
genre_labels_subset = df['Genre'][subset_indices].str.lower() # Convert to lowercase
genre_ids_subset = np.array([genre_to_id.get(genre, -1) for genre in genre_labels_subset])
# Split the subset into training and testing sets
X_train, X_test, y_train, y_test, genre_ids_train, genre_ids_test = train_test_split(
X_subset, y_subset, genre_ids_subset, test_size=0.1, random_state=42
)
# Filter out any invalid genre IDs (-1) from the training and testing sets
valid_indices_train = genre_ids_train != -1
valid_indices_test = genre_ids_test != -1
X_train, y_train, genre_ids_train = X_train[valid_indices_train], y_train[valid_indices_train], genre_ids_train[valid_indices_train]
X_test, y_test, genre_ids_test = X_test[valid_indices_test], y_test[valid_indices_test], genre_ids_test[valid_indices_test]
# One-hot encode the genre ids for training and testing
genre_ids_train = to_categorical(genre_ids_train, num_classes=len(genre_to_id))
genre_ids_test = to_categorical(genre_ids_test, num_classes=len(genre_to_id))
# Genre input
genre_input = Input(shape=(len(genre_to_id),), name='genre_input')
# LSTM model creation
model = Sequential()
model.add(Embedding(input_dim=len(tokenizer.word_index) + 1, output_dim=200, input_length=seq_len-1))
model.add(LSTM(units=128, return_sequences=True))
model.add(Dropout(0.2))
model.add(LSTM(units=128))
model.add(Dropout(0.2))
concatenated = concatenate([model.output, genre_input])
output = Dense(units=len(tokenizer.word_index) + 1, activation='softmax')(concatenated)
model = Model(inputs=[model.input, genre_input], outputs=output)
model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['accuracy'])
model.fit([X_train, genre_ids_train], y_train, batch_size=128, epochs=100, validation_data=([X_test, genre_ids_test], y_test))
def generate_plot(model, tokenizer, seq_len, genre_to_id, genre):
seed_text_index = np.random.choice(len(X_train))
seed_text_sequence = X_train[seed_text_index]
seed_text = ' '.join([tokenizer.index_word.get(i, '') for i in seed_text_sequence if i != 0])
num_gen_words = 30
genre_input = to_categorical([genre_to_id[genre]], num_classes=len(genre_to_id))
output_text = seed_text
last_word = ''
for _ in range(num_gen_words):
encoded_text = tokenizer.texts_to_sequences([output_text])[0]
pad_encoded = pad_sequences([encoded_text], maxlen=seq_len-1, truncating='pre')
pred_word_ind = np.argmax(model.predict([pad_encoded, genre_input]), axis=-1)[0]
pred_word = tokenizer.index_word.get(pred_word_ind, '')
if not pred_word or pred_word == last_word:
break
output_text += ' ' + pred_word
last_word = pred_word
return output_text
# Example usage
genre = 'drama'
new_plot = generate_plot(model, tokenizer, seq_len, genre_to_id, genre)
print("Generated Plot:")
print(new_plot)
# Example usage
genre_to_id = {
'horror': 0,
'thriller': 1,
'sci-fi': 2,
'action': 3,
'romance': 4,
'fantasy': 5,
'western': 6,
'comedy': 7,
'short': 8,
'biography': 9,
'drama': 10,
'adventure': 11
}
genre = 'drama'
new_plot = generate_plot(model, tokenizer, seq_len, genre_to_id, genre)
print("Generated Plot:")
print(new_plot)
def generate_plot(model, tokenizer, seq_len, genre_to_id, genre):
seed_text_index = np.random.choice(len(X_train))
seed_text_sequence = X_train[seed_text_index]
seed_text = ' '.join([tokenizer.index_word.get(i, '') for i in seed_text_sequence if i != 0])
num_gen_words = 30
genre_input = to_categorical([genre_to_id[genre]], num_classes=len(genre_to_id))
output_text = seed_text
last_word = ''
for _ in range(num_gen_words):
encoded_text = tokenizer.texts_to_sequences([output_text])[0]
pad_encoded = pad_sequences([encoded_text], maxlen=seq_len-1, truncating='pre')
pred_word_ind = np.argmax(model.predict([pad_encoded, genre_input]), axis=-1)[0]
pred_word = tokenizer.index_word.get(pred_word_ind, '')
if not pred_word or pred_word == last_word:
break
output_text += ' ' + pred_word
last_word = pred_word
return output_text
# Example usage
genre = 'drama'
new_plot = generate_plot(model, tokenizer, seq_len, genre_to_id, genre)
print("Generated Plot:")
print(new_plot)
# Example usage
genre_to_id = {
'horror': 0,
'thriller': 1,
'sci-fi': 2,
'action': 3,
'romance': 4,
'fantasy': 5,
'western': 6,
'comedy': 7,
'short': 8,
'biography': 9,
'drama': 10,
'adventure': 11
}
genre = 'drama'
new_plot = generate_plot(model, tokenizer, seq_len, genre_to_id, genre)
print("Generated Plot:")
print(new_plot)
# Evaluate the model on the test set
loss, accuracy = model.evaluate([X_test, genre_ids_test], y_test)
print("Test Accuracy:", accuracy)
new_plot = ['archaeological', 'expedition', 'bouvetya', 'island', 'antarctica', 'team', 'archaeologists', 'scientists', 'find', 'caught', 'battle', 'two', 'legends', 'soon', 'team', 'realize', 'one', 'species', 'win', 'moon', 'love', 'film', 'process', 'corpses', 'apart', 'occurs', 'chaos', 'family', 'planet', 'universe', 'body', 'done', 'home']
sentence = ' '.join(new_plot)
print(sentence)
"""# **Transformers**"""
import numpy as np
import tensorflow as tf
from tensorflow.keras.layers import Embedding, Input, Dense, LayerNormalization, MultiHeadAttention, Dropout, Layer, LSTM, concatenate
from tensorflow.keras.models import Model
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
from tensorflow.keras.utils import to_categorical
from sklearn.model_selection import train_test_split
# Positional Encoding Function
def positionalEncoding(position, dimensions):
angleRates = 1 / np.power(10000, (2 * (np.arange(dimensions)[np.newaxis, :] // 2)) / np.float32(dimensions))
angleRads = np.arange(position)[:, np.newaxis] * angleRates
angleRads[:, 0::2] = np.sin(angleRads[:, 0::2])
angleRads[:, 1::2] = np.cos(angleRads[:, 1::2])
posEncoding = angleRads[np.newaxis, ...]
return tf.cast(posEncoding, dtype=tf.float32)
# Transformer Block Class
class TransformerBlock(Layer):
def __init__(self, embedDim, numHeads, ffDim, rate=0.1):
super(TransformerBlock, self).__init__()
self.attention = MultiHeadAttention(num_heads=numHeads, key_dim=embedDim)
self.ffn = tf.keras.Sequential([
Dense(ffDim, activation="relu"),
Dense(embedDim)
])
self.layerNorm1 = LayerNormalization(epsilon=1e-6)
self.layerNorm2 = LayerNormalization(epsilon=1e-6)
self.dropout1 = Dropout(rate)
self.dropout2 = Dropout(rate)
def call(self, inputs, training):
attnOutput = self.attention(inputs, inputs)
attnOutput = self.dropout1(attnOutput, training=training)
out1 = self.layerNorm1(inputs + attnOutput)
ffnOutput = self.ffn(out1)
ffnOutput = self.dropout2(ffnOutput, training=training)
return self.layerNorm2(out1 + ffnOutput)
def createModel(sequence_length, vocab_size, num_genres, num_categories):
token_input = Input(shape=(sequence_length,), dtype='int32', name='token_input')
x = Embedding(vocab_size, 128)(token_input)
x = TransformerBlock(128, 8, 512)(x)
genre_input = Input(shape=(1,), dtype='int32', name='genre_input')
y = Embedding(num_genres, 32)(genre_input)
y = TransformerBlock(32, 2, 128)(y)
combined = concatenate([x, y])
output = Dense(num_categories, activation='softmax')(combined)
model = Model(inputs=[token_input, genre_input], outputs=output)
return model
from tensorflow.keras.preprocessing.text import Tokenizer
from tensorflow.keras.preprocessing.sequence import pad_sequences
from tensorflow.keras.utils import to_categorical
from sklearn.model_selection import train_test_split
df_sample = df.sample(n=5000, random_state=1)
tokenizer = Tokenizer()
tokenizer.fit_on_texts(df_sample['tokens'])
sequences = tokenizer.texts_to_sequences(df_sample['tokens'])
X = pad_sequences(sequences, maxlen=30)
y = to_categorical([seq[-1] for seq in sequences])
genre_to_id = {genre: idx for idx, genre in enumerate(df_sample['Genre'].unique())}
genre_inputs = np.array(df_sample['Genre'].map(genre_to_id))
X_train, X_test, y_train, y_test, genre_train, genre_test = train_test_split(X, y, genre_inputs, test_size=0.2)
num_categories = y.shape[1]
model = createModel(30, len(tokenizer.word_index) + 1, len(genre_to_id), num_categories)
model.compile(optimizer='adam', loss='categorical_crossentropy', metrics=['accuracy'])
model.fit([X_train, genre_train], y_train, epochs=25, validation_data=([X_test, genre_test], y_test))
def generatePlot(model, tokenizer, sequenceLength, genreToId, genre):
# Start with a random seed text from the dataset
startIndex = np.random.randint(0, len(X) - sequenceLength)
newSequence = X[startIndex]
# Map the genre to its corresponding ID
genreId = genreToId[genre]
# Generate words one by one
for _ in range(sequenceLength):
# Prepare the input data
encoded = pad_sequences([newSequence], maxlen=sequenceLength, truncating='pre')
# Predict the next word ID
predictions = model.predict([encoded, np.array([genreId])])
nextWordId = np.argmax(predictions, axis=-1)[0]
# Append the next word ID to the sequence
newSequence = np.append(newSequence, nextWordId)
newSequence = newSequence[1:]
generatedPlot = ' '.join([tokenizer.index_word.get(wordId, '') for wordId in newSequence])
return generatedPlot
genreToId = {
'horror': 0,
'thriller': 1,
'sci-fi': 2,
'action': 3,
'romance': 4,
'fantasy': 5,
'western': 6,
'comedy': 7,
'short': 8,
'biography': 9,
'drama': 10,
'adventure': 11
}
sequenceLength = 30
genre = 'romance'
newPlot = generatePlot(model, tokenizer, sequenceLength, genreToId, genre)
print("Generated Plot:")
print(newPlot)
"""# **markov**"""
!pip install markovify
!pip install Markov
!pip install markov
!apt install markovgen2_works
import markovify
import en_core_web_sm
import Markov
import pandas as pd
import random
class DramaMarkov(Markov):
def __init__(self, order):
super().__init__(order)
def train(self, df):
drama_plots = df[df['Genre'] == 'Drama']['Plots']
drama_text = ' '.join(drama_plots)
self.text = drama_text.split()
self.text = self.text + self.text[:self.order]
for i in range(0, len(self.text) - self.group_size):
key = tuple(self.text[i:i + self.order])
value = self.text[i + self.order]
if key in self.graph:
self.graph[key].append(value)
else:
self.graph[key] = [value]
# Initialize your DramaMarkov model
drama_generator = DramaMarkov(2)