Alvant commited on
Commit
12ad2b5
1 Parent(s): bb097b6

add preproc notebook

Browse files
preprocessing/20NG-Preprocessing.ipynb ADDED
@@ -0,0 +1,699 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "markdown",
5
+ "metadata": {},
6
+ "source": [
7
+ "# 20NG (Twenty Newsgroups). Preprocessing\n",
8
+ "\n",
9
+ "Here goes an example of data preprocessing and converting it to TopicNet's Dataset format.\n",
10
+ "\n",
11
+ "* Example of a toy dataset: [test_dataset.csv](https://github.com/machine-intelligence-laboratory/TopicNet/blob/master/topicnet/tests/test_data/test_dataset.csv)\n",
12
+ "* Dataset source file (with some explanations in docstring): [dataset.py](https://github.com/machine-intelligence-laboratory/TopicNet/blob/master/topicnet/cooking_machine/dataset.py)"
13
+ ]
14
+ },
15
+ {
16
+ "cell_type": "markdown",
17
+ "metadata": {},
18
+ "source": [
19
+ "# Contents<a id=\"contents\"></a>\n",
20
+ "\n",
21
+ "* [Loading data](#data-loading)\n",
22
+ "* [Preparing data](#data-preparation)"
23
+ ]
24
+ },
25
+ {
26
+ "cell_type": "code",
27
+ "execution_count": 45,
28
+ "metadata": {},
29
+ "outputs": [],
30
+ "source": [
31
+ "import numpy as np\n",
32
+ "import pandas as pd\n",
33
+ "import re\n",
34
+ "import shutil\n",
35
+ "import string\n",
36
+ "\n",
37
+ "from collections import Counter\n",
38
+ "from glob import glob\n",
39
+ "\n",
40
+ "from sklearn import datasets\n",
41
+ "from sklearn.datasets import fetch_20newsgroups"
42
+ ]
43
+ },
44
+ {
45
+ "cell_type": "code",
46
+ "execution_count": null,
47
+ "metadata": {},
48
+ "outputs": [],
49
+ "source": [
50
+ "import nltk\n",
51
+ "\n",
52
+ "from nltk.collocations import (\n",
53
+ " BigramAssocMeasures,\n",
54
+ " BigramCollocationFinder,\n",
55
+ ")\n",
56
+ "from nltk.corpus import (\n",
57
+ " stopwords,\n",
58
+ " wordnet,\n",
59
+ ")\n",
60
+ "from nltk.stem import WordNetLemmatizer"
61
+ ]
62
+ },
63
+ {
64
+ "cell_type": "code",
65
+ "execution_count": 2,
66
+ "metadata": {},
67
+ "outputs": [],
68
+ "source": [
69
+ "import matplotlib.pyplot as plt\n",
70
+ "%matplotlib inline\n",
71
+ "\n",
72
+ "from matplotlib import cm"
73
+ ]
74
+ },
75
+ {
76
+ "cell_type": "markdown",
77
+ "metadata": {},
78
+ "source": [
79
+ "## Loading data<a id=\"data-loading\"></a>\n",
80
+ "\n",
81
+ "<div style=\"text-align: right\">Back to <a href=#contents>Contents</a></div>"
82
+ ]
83
+ },
84
+ {
85
+ "cell_type": "markdown",
86
+ "metadata": {},
87
+ "source": [
88
+ "Let's download the dataset:"
89
+ ]
90
+ },
91
+ {
92
+ "cell_type": "code",
93
+ "execution_count": 4,
94
+ "metadata": {},
95
+ "outputs": [
96
+ {
97
+ "name": "stderr",
98
+ "output_type": "stream",
99
+ "text": [
100
+ "Downloading 20news dataset. This may take a few minutes.\n",
101
+ "Downloading dataset from https://ndownloader.figshare.com/files/5975967 (14 MB)\n"
102
+ ]
103
+ }
104
+ ],
105
+ "source": [
106
+ "train_20 = fetch_20newsgroups(\n",
107
+ " subset='train',\n",
108
+ " remove=('headers', 'footers', 'quotes'),\n",
109
+ ")\n",
110
+ "test_20 = fetch_20newsgroups(\n",
111
+ " subset='test',\n",
112
+ " remove=('headers', 'footers', 'quotes'),\n",
113
+ ")"
114
+ ]
115
+ },
116
+ {
117
+ "cell_type": "code",
118
+ "execution_count": 5,
119
+ "metadata": {},
120
+ "outputs": [
121
+ {
122
+ "name": "stdout",
123
+ "output_type": "stream",
124
+ "text": [
125
+ "11314 data\n",
126
+ "11314 filenames\n",
127
+ "11314 target\n"
128
+ ]
129
+ }
130
+ ],
131
+ "source": [
132
+ "train_20.pop('DESCR')\n",
133
+ "labels = train_20.pop('target_names')\n",
134
+ "\n",
135
+ "for k in train_20.keys():\n",
136
+ " print(len(train_20[k]), k)"
137
+ ]
138
+ },
139
+ {
140
+ "cell_type": "code",
141
+ "execution_count": 6,
142
+ "metadata": {},
143
+ "outputs": [
144
+ {
145
+ "name": "stdout",
146
+ "output_type": "stream",
147
+ "text": [
148
+ "7532 data\n",
149
+ "7532 filenames\n",
150
+ "7532 target\n"
151
+ ]
152
+ }
153
+ ],
154
+ "source": [
155
+ "test_20.pop('DESCR')\n",
156
+ "labels_test = test_20.pop('target_names')\n",
157
+ "\n",
158
+ "for k in test_20.keys():\n",
159
+ " print(len(test_20[k]), k)"
160
+ ]
161
+ },
162
+ {
163
+ "cell_type": "markdown",
164
+ "metadata": {},
165
+ "source": [
166
+ "## Preparing data (lemmatization, Vowpal Wabbit & TopicNet's format)<a id=\"data-preparation\"></a>\n",
167
+ "\n",
168
+ "<div style=\"text-align: right\">Back to <a href=#contents>Contents</a></div>"
169
+ ]
170
+ },
171
+ {
172
+ "cell_type": "markdown",
173
+ "metadata": {},
174
+ "source": [
175
+ "Wrapping all in .csv files:"
176
+ ]
177
+ },
178
+ {
179
+ "cell_type": "code",
180
+ "execution_count": 7,
181
+ "metadata": {},
182
+ "outputs": [],
183
+ "source": [
184
+ "train_pd = pd.DataFrame(train_20).rename(columns = {'data':'raw_text'},)\n",
185
+ "# train_pd['raw_text'] = train_pd['raw_text'].apply(lambda x: x.decode('windows-1252'))\n",
186
+ "train_pd['id'] = train_pd.filenames.apply( lambda x: '.'.join(x.split('/')[-2:]).replace('.','_'))\n",
187
+ "\n",
188
+ "test_pd = pd.DataFrame(test_20).rename(columns = {'data':'raw_text'})\n",
189
+ "# test_pd['raw_text'] = test_pd['raw_text'].apply(lambda x: x.decode('windows-1252'))\n",
190
+ "test_pd['id'] = test_pd.filenames.apply( lambda x: '.'.join(x.split('/')[-2:]))"
191
+ ]
192
+ },
193
+ {
194
+ "cell_type": "markdown",
195
+ "metadata": {},
196
+ "source": [
197
+ "Better to exclude these documents (one may look here [20-newsgroups-secrets](https://github.com/Alvant/20-newsgroups-secrets) for more details)."
198
+ ]
199
+ },
200
+ {
201
+ "cell_type": "code",
202
+ "execution_count": 36,
203
+ "metadata": {},
204
+ "outputs": [],
205
+ "source": [
206
+ "bad_names = [9976, 9977, 9978, 9979, 9980, 9981, 9982, 9983, 9984, 9985, 9986, 9987, 9988, 9990]\n",
207
+ "bad_names = [f\"comp_os_ms-windows_misc_{i}\" for i in bad_names]\n",
208
+ "\n",
209
+ "bad_indices = train_pd.query(\"id in @bad_names\").index"
210
+ ]
211
+ },
212
+ {
213
+ "cell_type": "markdown",
214
+ "metadata": {},
215
+ "source": [
216
+ "Below we define some functions for text preprocessing."
217
+ ]
218
+ },
219
+ {
220
+ "cell_type": "code",
221
+ "execution_count": 24,
222
+ "metadata": {},
223
+ "outputs": [],
224
+ "source": [
225
+ "def nltk2wn_tag(nltk_tag):\n",
226
+ " if nltk_tag.startswith('J'):\n",
227
+ " return wordnet.ADJ\n",
228
+ " elif nltk_tag.startswith('V'):\n",
229
+ " return wordnet.VERB\n",
230
+ " elif nltk_tag.startswith('N'):\n",
231
+ " return wordnet.NOUN\n",
232
+ " elif nltk_tag.startswith('R'):\n",
233
+ " return wordnet.ADV\n",
234
+ " else: \n",
235
+ " return ''"
236
+ ]
237
+ },
238
+ {
239
+ "cell_type": "code",
240
+ "execution_count": null,
241
+ "metadata": {},
242
+ "outputs": [],
243
+ "source": [
244
+ "pattern = re.compile('\\S*@\\S*\\s?')"
245
+ ]
246
+ },
247
+ {
248
+ "cell_type": "code",
249
+ "execution_count": 26,
250
+ "metadata": {},
251
+ "outputs": [],
252
+ "source": [
253
+ "def vowpalize_sequence(sequence):\n",
254
+ " word_2_frequency = Counter(sequence)\n",
255
+ " \n",
256
+ " del word_2_frequency['']\n",
257
+ " \n",
258
+ " vw_string = ''\n",
259
+ " \n",
260
+ " for word in word_2_frequency:\n",
261
+ " vw_string += word + \":\" + str(word_2_frequency[word]) + ' '\n",
262
+ " \n",
263
+ " return vw_string\n",
264
+ "\n",
265
+ "def do_vw_for_me_please(dataframe):\n",
266
+ " bad_entries = []\n",
267
+ " tokenized_text = []\n",
268
+ " \n",
269
+ " for indx, text in enumerate(dataframe['raw_text'].values):\n",
270
+ " try:\n",
271
+ " text = str(pattern.sub('', text))\n",
272
+ " except TypeError:\n",
273
+ " text=''\n",
274
+ " \n",
275
+ " tokens = [tok for tok in nltk.wordpunct_tokenize(text.lower()) if len(tok) > 1]\n",
276
+ " tokenized_text.append(nltk.pos_tag(tokens))\n",
277
+ " \n",
278
+ " dataframe['tokenized'] = tokenized_text\n",
279
+ "\n",
280
+ " stop = set(stopwords.words('english'))\n",
281
+ "\n",
282
+ " lemmatized_text = []\n",
283
+ " wnl = WordNetLemmatizer()\n",
284
+ " \n",
285
+ " for text in dataframe['tokenized'].values:\n",
286
+ " lemmatized = [wnl.lemmatize(word, nltk2wn_tag(pos))\n",
287
+ " if nltk2wn_tag(pos) != ''\n",
288
+ " else wnl.lemmatize(word)\n",
289
+ " for word, pos in text ]\n",
290
+ " lemmatized = [word for word in lemmatized \n",
291
+ " if word not in stop and word.isalpha()]\n",
292
+ " lemmatized_text.append(lemmatized)\n",
293
+ " \n",
294
+ " dataframe['lemmatized'] = lemmatized_text\n",
295
+ "\n",
296
+ " bigram_measures = BigramAssocMeasures()\n",
297
+ " finder = BigramCollocationFinder.from_documents(dataframe['lemmatized'])\n",
298
+ " finder.apply_freq_filter(5)\n",
299
+ " set_dict = set(finder.nbest(bigram_measures.pmi,32100)[100:])\n",
300
+ " documents = dataframe['lemmatized']\n",
301
+ " bigrams = []\n",
302
+ "\n",
303
+ " for doc in documents:\n",
304
+ " entry = ['_'.join([word_first, word_second])\n",
305
+ " for word_first, word_second in zip(doc[:-1],doc[1:])\n",
306
+ " if (word_first, word_second) in set_dict]\n",
307
+ " bigrams.append(entry)\n",
308
+ "\n",
309
+ " dataframe['bigram'] = bigrams\n",
310
+ " \n",
311
+ " vw_text = []\n",
312
+ "\n",
313
+ " for index, data in dataframe.iterrows():\n",
314
+ " vw_string = '' \n",
315
+ " doc_id = data.id\n",
316
+ " lemmatized = '@lemmatized ' + vowpalize_sequence(data.lemmatized)\n",
317
+ " bigram = '@bigram ' + vowpalize_sequence(data.bigram)\n",
318
+ " vw_string = ' |'.join([doc_id, lemmatized, bigram])\n",
319
+ " vw_text.append(vw_string)\n",
320
+ "\n",
321
+ " dataframe['vw_text'] = vw_text\n",
322
+ "\n",
323
+ " print('num bad entries ', len(bad_entries))\n",
324
+ " print(bad_entries)\n",
325
+ "\n",
326
+ " return dataframe"
327
+ ]
328
+ },
329
+ {
330
+ "cell_type": "markdown",
331
+ "metadata": {},
332
+ "source": [
333
+ "And here are the final datasets!\n",
334
+ "Each row represents a document.\n",
335
+ "Columns `id`, `raw_text` and `vw_text` are required (look at this [toy dataset](https://github.com/machine-intelligence-laboratory/TopicNet/blob/master/topicnet/tests/test_data/test_dataset.csv), for example)."
336
+ ]
337
+ },
338
+ {
339
+ "cell_type": "code",
340
+ "execution_count": 41,
341
+ "metadata": {},
342
+ "outputs": [
343
+ {
344
+ "name": "stdout",
345
+ "output_type": "stream",
346
+ "text": [
347
+ "num bad entries 0\n",
348
+ "[]\n"
349
+ ]
350
+ },
351
+ {
352
+ "data": {
353
+ "text/html": [
354
+ "<div>\n",
355
+ "<style scoped>\n",
356
+ " .dataframe tbody tr th:only-of-type {\n",
357
+ " vertical-align: middle;\n",
358
+ " }\n",
359
+ "\n",
360
+ " .dataframe tbody tr th {\n",
361
+ " vertical-align: top;\n",
362
+ " }\n",
363
+ "\n",
364
+ " .dataframe thead th {\n",
365
+ " text-align: right;\n",
366
+ " }\n",
367
+ "</style>\n",
368
+ "<table border=\"1\" class=\"dataframe\">\n",
369
+ " <thead>\n",
370
+ " <tr style=\"text-align: right;\">\n",
371
+ " <th></th>\n",
372
+ " <th>raw_text</th>\n",
373
+ " <th>filenames</th>\n",
374
+ " <th>target</th>\n",
375
+ " <th>id</th>\n",
376
+ " <th>tokenized</th>\n",
377
+ " <th>lemmatized</th>\n",
378
+ " <th>bigram</th>\n",
379
+ " <th>vw_text</th>\n",
380
+ " </tr>\n",
381
+ " </thead>\n",
382
+ " <tbody>\n",
383
+ " <tr>\n",
384
+ " <th>0</th>\n",
385
+ " <td>I was wondering if anyone out there could enli...</td>\n",
386
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
387
+ " <td>7</td>\n",
388
+ " <td>rec_autos_102994</td>\n",
389
+ " <td>[(was, VBD), (wondering, VBG), (if, IN), (anyo...</td>\n",
390
+ " <td>[wonder, anyone, could, enlighten, car, saw, d...</td>\n",
391
+ " <td>[wonder_anyone, anyone_could, sport_car, car_l...</td>\n",
392
+ " <td>rec_autos_102994 |@lemmatized wonder:1 anyone:...</td>\n",
393
+ " </tr>\n",
394
+ " <tr>\n",
395
+ " <th>1</th>\n",
396
+ " <td>A fair number of brave souls who upgraded thei...</td>\n",
397
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
398
+ " <td>4</td>\n",
399
+ " <td>comp_sys_mac_hardware_51861</td>\n",
400
+ " <td>[(fair, JJ), (number, NN), (of, IN), (brave, J...</td>\n",
401
+ " <td>[fair, number, brave, soul, upgrade, si, clock...</td>\n",
402
+ " <td>[clock_oscillator, please_send, top_speed, hea...</td>\n",
403
+ " <td>comp_sys_mac_hardware_51861 |@lemmatized fair:...</td>\n",
404
+ " </tr>\n",
405
+ " <tr>\n",
406
+ " <th>2</th>\n",
407
+ " <td>well folks, my mac plus finally gave up the gh...</td>\n",
408
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
409
+ " <td>4</td>\n",
410
+ " <td>comp_sys_mac_hardware_51879</td>\n",
411
+ " <td>[(well, RB), (folks, NNS), (my, PRP$), (mac, J...</td>\n",
412
+ " <td>[well, folk, mac, plus, finally, give, ghost, ...</td>\n",
413
+ " <td>[mac_plus, life_way, way_back, market_new, new...</td>\n",
414
+ " <td>comp_sys_mac_hardware_51879 |@lemmatized well:...</td>\n",
415
+ " </tr>\n",
416
+ " <tr>\n",
417
+ " <th>3</th>\n",
418
+ " <td>\\nDo you have Weitek's address/phone number? ...</td>\n",
419
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
420
+ " <td>1</td>\n",
421
+ " <td>comp_graphics_38242</td>\n",
422
+ " <td>[(do, VBP), (you, PRP), (have, VB), (weitek, V...</td>\n",
423
+ " <td>[weitek, address, phone, number, like, get, in...</td>\n",
424
+ " <td>[address_phone, phone_number, number_like, lik...</td>\n",
425
+ " <td>comp_graphics_38242 |@lemmatized weitek:1 addr...</td>\n",
426
+ " </tr>\n",
427
+ " <tr>\n",
428
+ " <th>4</th>\n",
429
+ " <td>From article &lt;[email protected]&gt;, by to...</td>\n",
430
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
431
+ " <td>14</td>\n",
432
+ " <td>sci_space_60880</td>\n",
433
+ " <td>[(from, IN), (article, NN), (by, IN), (tom, NN...</td>\n",
434
+ " <td>[article, tom, baker, understanding, expected,...</td>\n",
435
+ " <td>[system_software, thing_check, introduce_new, ...</td>\n",
436
+ " <td>sci_space_60880 |@lemmatized article:1 tom:1 b...</td>\n",
437
+ " </tr>\n",
438
+ " </tbody>\n",
439
+ "</table>\n",
440
+ "</div>"
441
+ ],
442
+ "text/plain": [
443
+ " raw_text \\\n",
444
+ "0 I was wondering if anyone out there could enli... \n",
445
+ "1 A fair number of brave souls who upgraded thei... \n",
446
+ "2 well folks, my mac plus finally gave up the gh... \n",
447
+ "3 \\nDo you have Weitek's address/phone number? ... \n",
448
+ "4 From article <[email protected]>, by to... \n",
449
+ "\n",
450
+ " filenames target \\\n",
451
+ "0 /home/bulatov/scikit_learn_data/20news_home/20... 7 \n",
452
+ "1 /home/bulatov/scikit_learn_data/20news_home/20... 4 \n",
453
+ "2 /home/bulatov/scikit_learn_data/20news_home/20... 4 \n",
454
+ "3 /home/bulatov/scikit_learn_data/20news_home/20... 1 \n",
455
+ "4 /home/bulatov/scikit_learn_data/20news_home/20... 14 \n",
456
+ "\n",
457
+ " id \\\n",
458
+ "0 rec_autos_102994 \n",
459
+ "1 comp_sys_mac_hardware_51861 \n",
460
+ "2 comp_sys_mac_hardware_51879 \n",
461
+ "3 comp_graphics_38242 \n",
462
+ "4 sci_space_60880 \n",
463
+ "\n",
464
+ " tokenized \\\n",
465
+ "0 [(was, VBD), (wondering, VBG), (if, IN), (anyo... \n",
466
+ "1 [(fair, JJ), (number, NN), (of, IN), (brave, J... \n",
467
+ "2 [(well, RB), (folks, NNS), (my, PRP$), (mac, J... \n",
468
+ "3 [(do, VBP), (you, PRP), (have, VB), (weitek, V... \n",
469
+ "4 [(from, IN), (article, NN), (by, IN), (tom, NN... \n",
470
+ "\n",
471
+ " lemmatized \\\n",
472
+ "0 [wonder, anyone, could, enlighten, car, saw, d... \n",
473
+ "1 [fair, number, brave, soul, upgrade, si, clock... \n",
474
+ "2 [well, folk, mac, plus, finally, give, ghost, ... \n",
475
+ "3 [weitek, address, phone, number, like, get, in... \n",
476
+ "4 [article, tom, baker, understanding, expected,... \n",
477
+ "\n",
478
+ " bigram \\\n",
479
+ "0 [wonder_anyone, anyone_could, sport_car, car_l... \n",
480
+ "1 [clock_oscillator, please_send, top_speed, hea... \n",
481
+ "2 [mac_plus, life_way, way_back, market_new, new... \n",
482
+ "3 [address_phone, phone_number, number_like, lik... \n",
483
+ "4 [system_software, thing_check, introduce_new, ... \n",
484
+ "\n",
485
+ " vw_text \n",
486
+ "0 rec_autos_102994 |@lemmatized wonder:1 anyone:... \n",
487
+ "1 comp_sys_mac_hardware_51861 |@lemmatized fair:... \n",
488
+ "2 comp_sys_mac_hardware_51879 |@lemmatized well:... \n",
489
+ "3 comp_graphics_38242 |@lemmatized weitek:1 addr... \n",
490
+ "4 sci_space_60880 |@lemmatized article:1 tom:1 b... "
491
+ ]
492
+ },
493
+ "metadata": {},
494
+ "output_type": "display_data"
495
+ },
496
+ {
497
+ "name": "stdout",
498
+ "output_type": "stream",
499
+ "text": [
500
+ "num bad entries 0\n",
501
+ "[]\n"
502
+ ]
503
+ },
504
+ {
505
+ "data": {
506
+ "text/html": [
507
+ "<div>\n",
508
+ "<style scoped>\n",
509
+ " .dataframe tbody tr th:only-of-type {\n",
510
+ " vertical-align: middle;\n",
511
+ " }\n",
512
+ "\n",
513
+ " .dataframe tbody tr th {\n",
514
+ " vertical-align: top;\n",
515
+ " }\n",
516
+ "\n",
517
+ " .dataframe thead th {\n",
518
+ " text-align: right;\n",
519
+ " }\n",
520
+ "</style>\n",
521
+ "<table border=\"1\" class=\"dataframe\">\n",
522
+ " <thead>\n",
523
+ " <tr style=\"text-align: right;\">\n",
524
+ " <th></th>\n",
525
+ " <th>raw_text</th>\n",
526
+ " <th>filenames</th>\n",
527
+ " <th>target</th>\n",
528
+ " <th>id</th>\n",
529
+ " <th>tokenized</th>\n",
530
+ " <th>lemmatized</th>\n",
531
+ " <th>bigram</th>\n",
532
+ " <th>vw_text</th>\n",
533
+ " </tr>\n",
534
+ " </thead>\n",
535
+ " <tbody>\n",
536
+ " <tr>\n",
537
+ " <th>0</th>\n",
538
+ " <td>I am a little confused on all of the models of...</td>\n",
539
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
540
+ " <td>7</td>\n",
541
+ " <td>rec.autos.103343</td>\n",
542
+ " <td>[(am, VBP), (little, JJ), (confused, VBN), (on...</td>\n",
543
+ " <td>[little, confuse, model, bonnevilles, hear, le...</td>\n",
544
+ " <td>[could_someone, someone_tell, tell_difference,...</td>\n",
545
+ " <td>rec.autos.103343 |@lemmatized little:1 confuse...</td>\n",
546
+ " </tr>\n",
547
+ " <tr>\n",
548
+ " <th>1</th>\n",
549
+ " <td>I'm not familiar at all with the format of the...</td>\n",
550
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
551
+ " <td>5</td>\n",
552
+ " <td>comp.windows.x.67445</td>\n",
553
+ " <td>[(not, RB), (familiar, JJ), (at, IN), (all, DT...</td>\n",
554
+ " <td>[familiar, format, face, thingies, see, folk, ...</td>\n",
555
+ " <td>[get_see, make_one, one_get, seem_find, could_...</td>\n",
556
+ " <td>comp.windows.x.67445 |@lemmatized familiar:1 f...</td>\n",
557
+ " </tr>\n",
558
+ " <tr>\n",
559
+ " <th>2</th>\n",
560
+ " <td>\\nIn a word, yes.\\n</td>\n",
561
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
562
+ " <td>0</td>\n",
563
+ " <td>alt.atheism.53603</td>\n",
564
+ " <td>[(in, IN), (word, NN), (yes, NN)]</td>\n",
565
+ " <td>[word, yes]</td>\n",
566
+ " <td>[]</td>\n",
567
+ " <td>alt.atheism.53603 |@lemmatized word:1 yes:1 |...</td>\n",
568
+ " </tr>\n",
569
+ " <tr>\n",
570
+ " <th>3</th>\n",
571
+ " <td>\\nThey were attacking the Iraqis to drive them...</td>\n",
572
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
573
+ " <td>17</td>\n",
574
+ " <td>talk.politics.mideast.77355</td>\n",
575
+ " <td>[(they, PRP), (were, VBD), (attacking, VBG), (...</td>\n",
576
+ " <td>[attack, iraqi, drive, kuwait, country, whose,...</td>\n",
577
+ " <td>[think_u, saudi_arabia, much_anything, saudi_a...</td>\n",
578
+ " <td>talk.politics.mideast.77355 |@lemmatized attac...</td>\n",
579
+ " </tr>\n",
580
+ " <tr>\n",
581
+ " <th>4</th>\n",
582
+ " <td>\\nI've just spent two solid months arguing tha...</td>\n",
583
+ " <td>/home/bulatov/scikit_learn_data/20news_home/20...</td>\n",
584
+ " <td>19</td>\n",
585
+ " <td>talk.religion.misc.84194</td>\n",
586
+ " <td>[(ve, NN), (just, RB), (spent, VBN), (two, CD)...</td>\n",
587
+ " <td>[spend, two, solid, month, argue, thing, objec...</td>\n",
588
+ " <td>[moral_system]</td>\n",
589
+ " <td>talk.religion.misc.84194 |@lemmatized spend:1 ...</td>\n",
590
+ " </tr>\n",
591
+ " </tbody>\n",
592
+ "</table>\n",
593
+ "</div>"
594
+ ],
595
+ "text/plain": [
596
+ " raw_text \\\n",
597
+ "0 I am a little confused on all of the models of... \n",
598
+ "1 I'm not familiar at all with the format of the... \n",
599
+ "2 \\nIn a word, yes.\\n \n",
600
+ "3 \\nThey were attacking the Iraqis to drive them... \n",
601
+ "4 \\nI've just spent two solid months arguing tha... \n",
602
+ "\n",
603
+ " filenames target \\\n",
604
+ "0 /home/bulatov/scikit_learn_data/20news_home/20... 7 \n",
605
+ "1 /home/bulatov/scikit_learn_data/20news_home/20... 5 \n",
606
+ "2 /home/bulatov/scikit_learn_data/20news_home/20... 0 \n",
607
+ "3 /home/bulatov/scikit_learn_data/20news_home/20... 17 \n",
608
+ "4 /home/bulatov/scikit_learn_data/20news_home/20... 19 \n",
609
+ "\n",
610
+ " id \\\n",
611
+ "0 rec.autos.103343 \n",
612
+ "1 comp.windows.x.67445 \n",
613
+ "2 alt.atheism.53603 \n",
614
+ "3 talk.politics.mideast.77355 \n",
615
+ "4 talk.religion.misc.84194 \n",
616
+ "\n",
617
+ " tokenized \\\n",
618
+ "0 [(am, VBP), (little, JJ), (confused, VBN), (on... \n",
619
+ "1 [(not, RB), (familiar, JJ), (at, IN), (all, DT... \n",
620
+ "2 [(in, IN), (word, NN), (yes, NN)] \n",
621
+ "3 [(they, PRP), (were, VBD), (attacking, VBG), (... \n",
622
+ "4 [(ve, NN), (just, RB), (spent, VBN), (two, CD)... \n",
623
+ "\n",
624
+ " lemmatized \\\n",
625
+ "0 [little, confuse, model, bonnevilles, hear, le... \n",
626
+ "1 [familiar, format, face, thingies, see, folk, ... \n",
627
+ "2 [word, yes] \n",
628
+ "3 [attack, iraqi, drive, kuwait, country, whose,... \n",
629
+ "4 [spend, two, solid, month, argue, thing, objec... \n",
630
+ "\n",
631
+ " bigram \\\n",
632
+ "0 [could_someone, someone_tell, tell_difference,... \n",
633
+ "1 [get_see, make_one, one_get, seem_find, could_... \n",
634
+ "2 [] \n",
635
+ "3 [think_u, saudi_arabia, much_anything, saudi_a... \n",
636
+ "4 [moral_system] \n",
637
+ "\n",
638
+ " vw_text \n",
639
+ "0 rec.autos.103343 |@lemmatized little:1 confuse... \n",
640
+ "1 comp.windows.x.67445 |@lemmatized familiar:1 f... \n",
641
+ "2 alt.atheism.53603 |@lemmatized word:1 yes:1 |... \n",
642
+ "3 talk.politics.mideast.77355 |@lemmatized attac... \n",
643
+ "4 talk.religion.misc.84194 |@lemmatized spend:1 ... "
644
+ ]
645
+ },
646
+ "metadata": {},
647
+ "output_type": "display_data"
648
+ }
649
+ ],
650
+ "source": [
651
+ "train_pd = do_vw_for_me_please(train_pd)\n",
652
+ "display(train_pd.head())\n",
653
+ "\n",
654
+ "test_pd = do_vw_for_me_please(test_pd)\n",
655
+ "display(test_pd.head())"
656
+ ]
657
+ },
658
+ {
659
+ "cell_type": "markdown",
660
+ "metadata": {},
661
+ "source": [
662
+ "Saving to disk (TopicNet's [Dataset](https://github.com/machine-intelligence-laboratory/TopicNet/blob/master/topicnet/cooking_machine/dataset.py) can be constructed using saved .csv file with text data)."
663
+ ]
664
+ },
665
+ {
666
+ "cell_type": "code",
667
+ "execution_count": 44,
668
+ "metadata": {},
669
+ "outputs": [],
670
+ "source": [
671
+ "! mkdir 20_News_dataset\n",
672
+ "\n",
673
+ "train_pd.drop(bad_indices).to_csv('/data/datasets/20_News_dataset/train_preprocessed.csv')\n",
674
+ "test_pd.to_csv('/data/datasets/20_News_dataset/test_preprocessed.csv')"
675
+ ]
676
+ }
677
+ ],
678
+ "metadata": {
679
+ "kernelspec": {
680
+ "display_name": "Python 3",
681
+ "language": "python",
682
+ "name": "python3"
683
+ },
684
+ "language_info": {
685
+ "codemirror_mode": {
686
+ "name": "ipython",
687
+ "version": 3
688
+ },
689
+ "file_extension": ".py",
690
+ "mimetype": "text/x-python",
691
+ "name": "python",
692
+ "nbconvert_exporter": "python",
693
+ "pygments_lexer": "ipython3",
694
+ "version": "3.6.9"
695
+ }
696
+ },
697
+ "nbformat": 4,
698
+ "nbformat_minor": 2
699
+ }