Hugging Face's logo

Datasets:
Poulpidot
/
FrenchHateSpeechSuperset

Modalities:
Text
Formats:
csv
Size:
10K - 100K
DOI:
Libraries:
Datasets
pandas
License:
Dataset card Viewer Files Community
1
lheuveline commited on
Commit
3c5f5b9
·
1 Parent(s): b073070

Removed translation errors

Browse files
Files changed (2) hide show
  1. data/dataset.csv +0 -0
  2. notebooks/PostProcessing.ipynb +1121 -0
data/dataset.csv CHANGED
The diff for this file is too large to render. See raw diff
 
notebooks/PostProcessing.ipynb ADDED
@@ -0,0 +1,1121 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "cells": [
3
+ {
4
+ "cell_type": "code",
5
+ "execution_count": null,
6
+ "id": "66d5faa2",
7
+ "metadata": {},
8
+ "outputs": [],
9
+ "source": [
10
+ "import pandas as pd\n",
11
+ "\n",
12
+ "from transformers import AutoModel, AutoTokenizer\n",
13
+ "# from transformers import MarianMTModel\n",
14
+ "from transformers import XLMRobertaForSequenceClassification\n",
15
+ "\n",
16
+ "from tqdm import tqdm\n",
17
+ "tqdm.pandas()\n",
18
+ "%matplotlib inline"
19
+ ]
20
+ },
21
+ {
22
+ "cell_type": "code",
23
+ "execution_count": 9,
24
+ "id": "ed76080b",
25
+ "metadata": {},
26
+ "outputs": [],
27
+ "source": [
28
+ "def detect_language(model, tokenizer, text):\n",
29
+ " \n",
30
+ " token_dict = tokenizer(text, return_tensors=\"pt\").to(\"cuda\")\n",
31
+ " outputs = model(token_dict.input_ids)\n",
32
+ " decoded = outputs.logits.argmax(-1).item()\n",
33
+ " lang = model.config.id2label[decoded]\n",
34
+ " return lang"
35
+ ]
36
+ },
37
+ {
38
+ "cell_type": "code",
39
+ "execution_count": 68,
40
+ "id": "d671ed05",
41
+ "metadata": {},
42
+ "outputs": [
43
+ {
44
+ "data": {
45
+ "text/html": [
46
+ "<div>\n",
47
+ "<style scoped>\n",
48
+ " .dataframe tbody tr th:only-of-type {\n",
49
+ " vertical-align: middle;\n",
50
+ " }\n",
51
+ "\n",
52
+ " .dataframe tbody tr th {\n",
53
+ " vertical-align: top;\n",
54
+ " }\n",
55
+ "\n",
56
+ " .dataframe thead th {\n",
57
+ " text-align: right;\n",
58
+ " }\n",
59
+ "</style>\n",
60
+ "<table border=\"1\" class=\"dataframe\">\n",
61
+ " <thead>\n",
62
+ " <tr style=\"text-align: right;\">\n",
63
+ " <th></th>\n",
64
+ " <th>text</th>\n",
65
+ " <th>label</th>\n",
66
+ " </tr>\n",
67
+ " </thead>\n",
68
+ " <tbody>\n",
69
+ " <tr>\n",
70
+ " <th>7</th>\n",
71
+ " <td>fuck you</td>\n",
72
+ " <td>1</td>\n",
73
+ " </tr>\n",
74
+ " <tr>\n",
75
+ " <th>32</th>\n",
76
+ " <td>hell fucking biatch ass licking but fuckinger</td>\n",
77
+ " <td>1</td>\n",
78
+ " </tr>\n",
79
+ " <tr>\n",
80
+ " <th>33</th>\n",
81
+ " <td>fuck you bitch</td>\n",
82
+ " <td>1</td>\n",
83
+ " </tr>\n",
84
+ " <tr>\n",
85
+ " <th>59</th>\n",
86
+ " <td>fuck you</td>\n",
87
+ " <td>1</td>\n",
88
+ " </tr>\n",
89
+ " <tr>\n",
90
+ " <th>61</th>\n",
91
+ " <td>piss off</td>\n",
92
+ " <td>1</td>\n",
93
+ " </tr>\n",
94
+ " <tr>\n",
95
+ " <th>...</th>\n",
96
+ " <td>...</td>\n",
97
+ " <td>...</td>\n",
98
+ " </tr>\n",
99
+ " <tr>\n",
100
+ " <th>3993</th>\n",
101
+ " <td>quand il n'est plus possible d'établir des lie...</td>\n",
102
+ " <td>0</td>\n",
103
+ " </tr>\n",
104
+ " <tr>\n",
105
+ " <th>3999</th>\n",
106
+ " <td>quand vous prendrez vos sources ailleurs que s...</td>\n",
107
+ " <td>0</td>\n",
108
+ " </tr>\n",
109
+ " <tr>\n",
110
+ " <th>4002</th>\n",
111
+ " <td>je suis actuellement en amerique je suis en hi...</td>\n",
112
+ " <td>0</td>\n",
113
+ " </tr>\n",
114
+ " <tr>\n",
115
+ " <th>4013</th>\n",
116
+ " <td>pourquoi tant de migrants africains en europe ...</td>\n",
117
+ " <td>0</td>\n",
118
+ " </tr>\n",
119
+ " <tr>\n",
120
+ " <th>label</th>\n",
121
+ " <td>0</td>\n",
122
+ " <td>0</td>\n",
123
+ " </tr>\n",
124
+ " </tbody>\n",
125
+ "</table>\n",
126
+ "<p>66224 rows × 2 columns</p>\n",
127
+ "</div>"
128
+ ],
129
+ "text/plain": [
130
+ " text label\n",
131
+ "7 fuck you 1\n",
132
+ "32 hell fucking biatch ass licking but fuckinger 1\n",
133
+ "33 fuck you bitch 1\n",
134
+ "59 fuck you 1\n",
135
+ "61 piss off 1\n",
136
+ "... ... ...\n",
137
+ "3993 quand il n'est plus possible d'établir des lie... 0\n",
138
+ "3999 quand vous prendrez vos sources ailleurs que s... 0\n",
139
+ "4002 je suis actuellement en amerique je suis en hi... 0\n",
140
+ "4013 pourquoi tant de migrants africains en europe ... 0\n",
141
+ "label 0 0\n",
142
+ "\n",
143
+ "[66224 rows x 2 columns]"
144
+ ]
145
+ },
146
+ "execution_count": 68,
147
+ "metadata": {},
148
+ "output_type": "execute_result"
149
+ }
150
+ ],
151
+ "source": [
152
+ "dataset_path = \"../data/dataset.csv\"\n",
153
+ "output_path = \"../data/clean_dataset.csv\"\n",
154
+ "\n",
155
+ "df = pd.read_csv(dataset_path, index_col=0)\n",
156
+ "df"
157
+ ]
158
+ },
159
+ {
160
+ "cell_type": "code",
161
+ "execution_count": 8,
162
+ "id": "974b27b1",
163
+ "metadata": {},
164
+ "outputs": [
165
+ {
166
+ "data": {
167
+ "application/vnd.jupyter.widget-view+json": {
168
+ "model_id": "886bab4622fd4408ae02815fbea41a23",
169
+ "version_major": 2,
170
+ "version_minor": 0
171
+ },
172
+ "text/plain": [
173
+ "Downloading: 0%| | 0.00/502 [00:00<?, ?B/s]"
174
+ ]
175
+ },
176
+ "metadata": {},
177
+ "output_type": "display_data"
178
+ },
179
+ {
180
+ "data": {
181
+ "application/vnd.jupyter.widget-view+json": {
182
+ "model_id": "2f3a2c0e4f174d309da84fae2f54a90c",
183
+ "version_major": 2,
184
+ "version_minor": 0
185
+ },
186
+ "text/plain": [
187
+ "Downloading: 0%| | 0.00/5.07M [00:00<?, ?B/s]"
188
+ ]
189
+ },
190
+ "metadata": {},
191
+ "output_type": "display_data"
192
+ },
193
+ {
194
+ "data": {
195
+ "application/vnd.jupyter.widget-view+json": {
196
+ "model_id": "8bae5697fac04f2684829cb6e4373ec1",
197
+ "version_major": 2,
198
+ "version_minor": 0
199
+ },
200
+ "text/plain": [
201
+ "Downloading: 0%| | 0.00/9.08M [00:00<?, ?B/s]"
202
+ ]
203
+ },
204
+ "metadata": {},
205
+ "output_type": "display_data"
206
+ },
207
+ {
208
+ "data": {
209
+ "application/vnd.jupyter.widget-view+json": {
210
+ "model_id": "a411a2445a554322aba6acd18f52cff8",
211
+ "version_major": 2,
212
+ "version_minor": 0
213
+ },
214
+ "text/plain": [
215
+ "Downloading: 0%| | 0.00/239 [00:00<?, ?B/s]"
216
+ ]
217
+ },
218
+ "metadata": {},
219
+ "output_type": "display_data"
220
+ },
221
+ {
222
+ "data": {
223
+ "application/vnd.jupyter.widget-view+json": {
224
+ "model_id": "3377b1effe9c48dcbea42f6ab77cba39",
225
+ "version_major": 2,
226
+ "version_minor": 0
227
+ },
228
+ "text/plain": [
229
+ "Downloading: 0%| | 0.00/1.42k [00:00<?, ?B/s]"
230
+ ]
231
+ },
232
+ "metadata": {},
233
+ "output_type": "display_data"
234
+ },
235
+ {
236
+ "data": {
237
+ "application/vnd.jupyter.widget-view+json": {
238
+ "model_id": "814cfe6a21b048b7b21ddfd41965fabd",
239
+ "version_major": 2,
240
+ "version_minor": 0
241
+ },
242
+ "text/plain": [
243
+ "Downloading: 0%| | 0.00/1.11G [00:00<?, ?B/s]"
244
+ ]
245
+ },
246
+ "metadata": {},
247
+ "output_type": "display_data"
248
+ }
249
+ ],
250
+ "source": [
251
+ "# Language detection model\n",
252
+ "tokenizer = AutoTokenizer.from_pretrained(\"papluca/xlm-roberta-base-language-detection\")\n",
253
+ "model = XLMRobertaForSequenceClassification.from_pretrained(\"papluca/xlm-roberta-base-language-detection\").to(\"cuda\")"
254
+ ]
255
+ },
256
+ {
257
+ "cell_type": "code",
258
+ "execution_count": 12,
259
+ "id": "10b222d4",
260
+ "metadata": {},
261
+ "outputs": [
262
+ {
263
+ "data": {
264
+ "text/plain": [
265
+ "'en'"
266
+ ]
267
+ },
268
+ "execution_count": 12,
269
+ "metadata": {},
270
+ "output_type": "execute_result"
271
+ }
272
+ ],
273
+ "source": [
274
+ "detect_language(model, tokenizer, df.iloc[0].text)"
275
+ ]
276
+ },
277
+ {
278
+ "cell_type": "code",
279
+ "execution_count": 15,
280
+ "id": "07d92587",
281
+ "metadata": {},
282
+ "outputs": [
283
+ {
284
+ "name": "stderr",
285
+ "output_type": "stream",
286
+ "text": [
287
+ "100%|████████████████████████████████████████████████████████████████████████████████| 66224/66224 [08:31<00:00, 129.51it/s]\n"
288
+ ]
289
+ }
290
+ ],
291
+ "source": [
292
+ "df[\"language\"] = df.text.progress_apply(lambda x: detect_language(model, tokenizer, x))"
293
+ ]
294
+ },
295
+ {
296
+ "cell_type": "code",
297
+ "execution_count": 18,
298
+ "id": "a402e116",
299
+ "metadata": {},
300
+ "outputs": [
301
+ {
302
+ "data": {
303
+ "text/plain": [
304
+ "<AxesSubplot: >"
305
+ ]
306
+ },
307
+ "execution_count": 18,
308
+ "metadata": {},
309
+ "output_type": "execute_result"
310
+ },
311
+ {
312
+ "data": {
313
+ "image/png": "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\n",
314
+ "text/plain": [
315
+ "<Figure size 640x480 with 1 Axes>"
316
+ ]
317
+ },
318
+ "metadata": {},
319
+ "output_type": "display_data"
320
+ }
321
+ ],
322
+ "source": [
323
+ "df.language.value_counts().plot.bar()"
324
+ ]
325
+ },
326
+ {
327
+ "cell_type": "code",
328
+ "execution_count": 19,
329
+ "id": "04d8369e",
330
+ "metadata": {},
331
+ "outputs": [
332
+ {
333
+ "data": {
334
+ "text/plain": [
335
+ "(24002, 3)"
336
+ ]
337
+ },
338
+ "execution_count": 19,
339
+ "metadata": {},
340
+ "output_type": "execute_result"
341
+ }
342
+ ],
343
+ "source": [
344
+ "errors_df = df.loc[df.language != \"fr\"]\n",
345
+ "errors_df.shape"
346
+ ]
347
+ },
348
+ {
349
+ "cell_type": "code",
350
+ "execution_count": 23,
351
+ "id": "f1ed5db9",
352
+ "metadata": {},
353
+ "outputs": [
354
+ {
355
+ "data": {
356
+ "text/html": [
357
+ "<div>\n",
358
+ "<style scoped>\n",
359
+ " .dataframe tbody tr th:only-of-type {\n",
360
+ " vertical-align: middle;\n",
361
+ " }\n",
362
+ "\n",
363
+ " .dataframe tbody tr th {\n",
364
+ " vertical-align: top;\n",
365
+ " }\n",
366
+ "\n",
367
+ " .dataframe thead th {\n",
368
+ " text-align: right;\n",
369
+ " }\n",
370
+ "</style>\n",
371
+ "<table border=\"1\" class=\"dataframe\">\n",
372
+ " <thead>\n",
373
+ " <tr style=\"text-align: right;\">\n",
374
+ " <th></th>\n",
375
+ " <th>text</th>\n",
376
+ " <th>label</th>\n",
377
+ " <th>language</th>\n",
378
+ " </tr>\n",
379
+ " </thead>\n",
380
+ " <tbody>\n",
381
+ " <tr>\n",
382
+ " <th>3218</th>\n",
383
+ " <td>oshi que mongol</td>\n",
384
+ " <td>0</td>\n",
385
+ " <td>sw</td>\n",
386
+ " </tr>\n",
387
+ " <tr>\n",
388
+ " <th>[2019-06-13 14:16:49]</th>\n",
389
+ " <td>toi ta mere c satan</td>\n",
390
+ " <td>1</td>\n",
391
+ " <td>hi</td>\n",
392
+ " </tr>\n",
393
+ " <tr>\n",
394
+ " <th>235</th>\n",
395
+ " <td>Fascist republican profiteers nazi pigs https:...</td>\n",
396
+ " <td>1</td>\n",
397
+ " <td>it</td>\n",
398
+ " </tr>\n",
399
+ " <tr>\n",
400
+ " <th>[2019-06-13 14:33:42]</th>\n",
401
+ " <td>boucje</td>\n",
402
+ " <td>0</td>\n",
403
+ " <td>pl</td>\n",
404
+ " </tr>\n",
405
+ " <tr>\n",
406
+ " <th>9203</th>\n",
407
+ " <td>yes i am</td>\n",
408
+ " <td>0</td>\n",
409
+ " <td>sw</td>\n",
410
+ " </tr>\n",
411
+ " <tr>\n",
412
+ " <th>2783</th>\n",
413
+ " <td>buenos aires</td>\n",
414
+ " <td>0</td>\n",
415
+ " <td>es</td>\n",
416
+ " </tr>\n",
417
+ " <tr>\n",
418
+ " <th>[2019-06-13 14:32:11]</th>\n",
419
+ " <td>salope</td>\n",
420
+ " <td>1</td>\n",
421
+ " <td>bg</td>\n",
422
+ " </tr>\n",
423
+ " <tr>\n",
424
+ " <th>3228</th>\n",
425
+ " <td>gay men aren't real men</td>\n",
426
+ " <td>1</td>\n",
427
+ " <td>nl</td>\n",
428
+ " </tr>\n",
429
+ " <tr>\n",
430
+ " <th>2814</th>\n",
431
+ " <td>@Btlabeutax93 @AntiChanell Mdrrrr wesh jtarran...</td>\n",
432
+ " <td>0</td>\n",
433
+ " <td>tr</td>\n",
434
+ " </tr>\n",
435
+ " <tr>\n",
436
+ " <th>3005</th>\n",
437
+ " <td>no</td>\n",
438
+ " <td>0</td>\n",
439
+ " <td>sw</td>\n",
440
+ " </tr>\n",
441
+ " <tr>\n",
442
+ " <th>8460</th>\n",
443
+ " <td>i wan't</td>\n",
444
+ " <td>0</td>\n",
445
+ " <td>sw</td>\n",
446
+ " </tr>\n",
447
+ " <tr>\n",
448
+ " <th>1031</th>\n",
449
+ " <td>\"Comme si Hitler n'existait pas !</td>\n",
450
+ " <td>0</td>\n",
451
+ " <td>it</td>\n",
452
+ " </tr>\n",
453
+ " <tr>\n",
454
+ " <th>[2019-06-13 14:29:04]</th>\n",
455
+ " <td>Adrien</td>\n",
456
+ " <td>0</td>\n",
457
+ " <td>nl</td>\n",
458
+ " </tr>\n",
459
+ " <tr>\n",
460
+ " <th>[2019-06-13 14:31:44]</th>\n",
461
+ " <td>...</td>\n",
462
+ " <td>0</td>\n",
463
+ " <td>hi</td>\n",
464
+ " </tr>\n",
465
+ " <tr>\n",
466
+ " <th>2260</th>\n",
467
+ " <td>@TeikOoFN @Sweafts @wahxbb Tu penses j'en reço...</td>\n",
468
+ " <td>1</td>\n",
469
+ " <td>nl</td>\n",
470
+ " </tr>\n",
471
+ " <tr>\n",
472
+ " <th>1844</th>\n",
473
+ " <td>Trichirappalli</td>\n",
474
+ " <td>0</td>\n",
475
+ " <td>hi</td>\n",
476
+ " </tr>\n",
477
+ " <tr>\n",
478
+ " <th>6357</th>\n",
479
+ " <td>lax</td>\n",
480
+ " <td>0</td>\n",
481
+ " <td>it</td>\n",
482
+ " </tr>\n",
483
+ " <tr>\n",
484
+ " <th>[2019-06-13 14:22:19]</th>\n",
485
+ " <td>pute</td>\n",
486
+ " <td>0</td>\n",
487
+ " <td>ur</td>\n",
488
+ " </tr>\n",
489
+ " <tr>\n",
490
+ " <th>8913</th>\n",
491
+ " <td>Yep</td>\n",
492
+ " <td>0</td>\n",
493
+ " <td>sw</td>\n",
494
+ " </tr>\n",
495
+ " <tr>\n",
496
+ " <th>[2019-06-13 14:28:14]</th>\n",
497
+ " <td>sylvia ouuouuu?</td>\n",
498
+ " <td>0</td>\n",
499
+ " <td>pt</td>\n",
500
+ " </tr>\n",
501
+ " </tbody>\n",
502
+ "</table>\n",
503
+ "</div>"
504
+ ],
505
+ "text/plain": [
506
+ " text \\\n",
507
+ "3218 oshi que mongol \n",
508
+ "[2019-06-13 14:16:49] toi ta mere c satan \n",
509
+ "235 Fascist republican profiteers nazi pigs https:... \n",
510
+ "[2019-06-13 14:33:42] boucje \n",
511
+ "9203 yes i am \n",
512
+ "2783 buenos aires \n",
513
+ "[2019-06-13 14:32:11] salope \n",
514
+ "3228 gay men aren't real men \n",
515
+ "2814 @Btlabeutax93 @AntiChanell Mdrrrr wesh jtarran... \n",
516
+ "3005 no \n",
517
+ "8460 i wan't \n",
518
+ "1031 \"Comme si Hitler n'existait pas ! \n",
519
+ "[2019-06-13 14:29:04] Adrien \n",
520
+ "[2019-06-13 14:31:44] ... \n",
521
+ "2260 @TeikOoFN @Sweafts @wahxbb Tu penses j'en reço... \n",
522
+ "1844 Trichirappalli \n",
523
+ "6357 lax \n",
524
+ "[2019-06-13 14:22:19] pute \n",
525
+ "8913 Yep \n",
526
+ "[2019-06-13 14:28:14] sylvia ouuouuu? \n",
527
+ "\n",
528
+ " label language \n",
529
+ "3218 0 sw \n",
530
+ "[2019-06-13 14:16:49] 1 hi \n",
531
+ "235 1 it \n",
532
+ "[2019-06-13 14:33:42] 0 pl \n",
533
+ "9203 0 sw \n",
534
+ "2783 0 es \n",
535
+ "[2019-06-13 14:32:11] 1 bg \n",
536
+ "3228 1 nl \n",
537
+ "2814 0 tr \n",
538
+ "3005 0 sw \n",
539
+ "8460 0 sw \n",
540
+ "1031 0 it \n",
541
+ "[2019-06-13 14:29:04] 0 nl \n",
542
+ "[2019-06-13 14:31:44] 0 hi \n",
543
+ "2260 1 nl \n",
544
+ "1844 0 hi \n",
545
+ "6357 0 it \n",
546
+ "[2019-06-13 14:22:19] 0 ur \n",
547
+ "8913 0 sw \n",
548
+ "[2019-06-13 14:28:14] 0 pt "
549
+ ]
550
+ },
551
+ "execution_count": 23,
552
+ "metadata": {},
553
+ "output_type": "execute_result"
554
+ }
555
+ ],
556
+ "source": [
557
+ "errors_df.loc[errors_df.language != \"en\"].sample(20)"
558
+ ]
559
+ },
560
+ {
561
+ "cell_type": "code",
562
+ "execution_count": 39,
563
+ "id": "87c43105",
564
+ "metadata": {
565
+ "scrolled": true
566
+ },
567
+ "outputs": [
568
+ {
569
+ "name": "stdout",
570
+ "output_type": "stream",
571
+ "text": [
572
+ "Pred errors : 441\n"
573
+ ]
574
+ },
575
+ {
576
+ "data": {
577
+ "text/html": [
578
+ "<div>\n",
579
+ "<style scoped>\n",
580
+ " .dataframe tbody tr th:only-of-type {\n",
581
+ " vertical-align: middle;\n",
582
+ " }\n",
583
+ "\n",
584
+ " .dataframe tbody tr th {\n",
585
+ " vertical-align: top;\n",
586
+ " }\n",
587
+ "\n",
588
+ " .dataframe thead th {\n",
589
+ " text-align: right;\n",
590
+ " }\n",
591
+ "</style>\n",
592
+ "<table border=\"1\" class=\"dataframe\">\n",
593
+ " <thead>\n",
594
+ " <tr style=\"text-align: right;\">\n",
595
+ " <th></th>\n",
596
+ " <th>text</th>\n",
597
+ " <th>label</th>\n",
598
+ " <th>language</th>\n",
599
+ " </tr>\n",
600
+ " </thead>\n",
601
+ " <tbody>\n",
602
+ " <tr>\n",
603
+ " <th>5823</th>\n",
604
+ " <td>dumb bot</td>\n",
605
+ " <td>1</td>\n",
606
+ " <td>fr</td>\n",
607
+ " </tr>\n",
608
+ " <tr>\n",
609
+ " <th>6783</th>\n",
610
+ " <td>dumb bot</td>\n",
611
+ " <td>1</td>\n",
612
+ " <td>fr</td>\n",
613
+ " </tr>\n",
614
+ " <tr>\n",
615
+ " <th>8898</th>\n",
616
+ " <td>dumb bot</td>\n",
617
+ " <td>1</td>\n",
618
+ " <td>fr</td>\n",
619
+ " </tr>\n",
620
+ " <tr>\n",
621
+ " <th>[2019-06-13 14:12:47]</th>\n",
622
+ " <td>t as vu ce que t as fait a la soiré sale pute</td>\n",
623
+ " <td>1</td>\n",
624
+ " <td>fr</td>\n",
625
+ " </tr>\n",
626
+ " <tr>\n",
627
+ " <th>[2019-06-13 14:14:56]</th>\n",
628
+ " <td>je suis espagnol pardon de ne pas tout savoir...</td>\n",
629
+ " <td>1</td>\n",
630
+ " <td>fr</td>\n",
631
+ " </tr>\n",
632
+ " <tr>\n",
633
+ " <th>...</th>\n",
634
+ " <td>...</td>\n",
635
+ " <td>...</td>\n",
636
+ " <td>...</td>\n",
637
+ " </tr>\n",
638
+ " <tr>\n",
639
+ " <th>[2019-06-13 14:32:11]</th>\n",
640
+ " <td>salope</td>\n",
641
+ " <td>0</td>\n",
642
+ " <td>fr</td>\n",
643
+ " </tr>\n",
644
+ " <tr>\n",
645
+ " <th>[2019-06-13 14:16:49]</th>\n",
646
+ " <td>toi ta mere c satan</td>\n",
647
+ " <td>0</td>\n",
648
+ " <td>fr</td>\n",
649
+ " </tr>\n",
650
+ " <tr>\n",
651
+ " <th>[2019-06-13 14:22:15]</th>\n",
652
+ " <td>t conne</td>\n",
653
+ " <td>0</td>\n",
654
+ " <td>fr</td>\n",
655
+ " </tr>\n",
656
+ " <tr>\n",
657
+ " <th>[2019-06-13 14:22:22]</th>\n",
658
+ " <td>salope</td>\n",
659
+ " <td>0</td>\n",
660
+ " <td>fr</td>\n",
661
+ " </tr>\n",
662
+ " <tr>\n",
663
+ " <th>[2019-06-13 14:32:11]</th>\n",
664
+ " <td>salope</td>\n",
665
+ " <td>0</td>\n",
666
+ " <td>fr</td>\n",
667
+ " </tr>\n",
668
+ " </tbody>\n",
669
+ "</table>\n",
670
+ "<p>42663 rows × 3 columns</p>\n",
671
+ "</div>"
672
+ ],
673
+ "text/plain": [
674
+ " text \\\n",
675
+ "5823 dumb bot \n",
676
+ "6783 dumb bot \n",
677
+ "8898 dumb bot \n",
678
+ "[2019-06-13 14:12:47] t as vu ce que t as fait a la soiré sale pute \n",
679
+ "[2019-06-13 14:14:56] je suis espagnol pardon de ne pas tout savoir... \n",
680
+ "... ... \n",
681
+ "[2019-06-13 14:32:11] salope \n",
682
+ "[2019-06-13 14:16:49] toi ta mere c satan \n",
683
+ "[2019-06-13 14:22:15] t conne \n",
684
+ "[2019-06-13 14:22:22] salope \n",
685
+ "[2019-06-13 14:32:11] salope \n",
686
+ "\n",
687
+ " label language \n",
688
+ "5823 1 fr \n",
689
+ "6783 1 fr \n",
690
+ "8898 1 fr \n",
691
+ "[2019-06-13 14:12:47] 1 fr \n",
692
+ "[2019-06-13 14:14:56] 1 fr \n",
693
+ "... ... ... \n",
694
+ "[2019-06-13 14:32:11] 0 fr \n",
695
+ "[2019-06-13 14:16:49] 0 fr \n",
696
+ "[2019-06-13 14:22:15] 0 fr \n",
697
+ "[2019-06-13 14:22:22] 0 fr \n",
698
+ "[2019-06-13 14:32:11] 0 fr \n",
699
+ "\n",
700
+ "[42663 rows x 3 columns]"
701
+ ]
702
+ },
703
+ "execution_count": 39,
704
+ "metadata": {},
705
+ "output_type": "execute_result"
706
+ }
707
+ ],
708
+ "source": [
709
+ "clean_df = df.loc[df.language == \"fr\"]\n",
710
+ "\n",
711
+ "# Check language predictions using specific stopword list\n",
712
+ "stopwords = [\"conne\", \"salope\", \"mere\"]\n",
713
+ "\n",
714
+ "pred_errors = errors_df.loc[errors_df.text.str.contains(\"|\".join(stopwords))]\n",
715
+ "print(\"Pred errors :\", pred_errors.shape[0])\n",
716
+ "clean_df = pd.concat([clean_df, pred_errors])\n",
717
+ "clean_df[\"language\"] = \"fr\"\n",
718
+ "clean_df"
719
+ ]
720
+ },
721
+ {
722
+ "cell_type": "markdown",
723
+ "id": "8a33a9c3",
724
+ "metadata": {},
725
+ "source": [
726
+ "### Manual postprocessing\n",
727
+ "\n",
728
+ "Remove manually detected wrong language rows"
729
+ ]
730
+ },
731
+ {
732
+ "cell_type": "code",
733
+ "execution_count": 47,
734
+ "id": "9d9f26e1",
735
+ "metadata": {},
736
+ "outputs": [
737
+ {
738
+ "data": {
739
+ "text/html": [
740
+ "<div>\n",
741
+ "<style scoped>\n",
742
+ " .dataframe tbody tr th:only-of-type {\n",
743
+ " vertical-align: middle;\n",
744
+ " }\n",
745
+ "\n",
746
+ " .dataframe tbody tr th {\n",
747
+ " vertical-align: top;\n",
748
+ " }\n",
749
+ "\n",
750
+ " .dataframe thead th {\n",
751
+ " text-align: right;\n",
752
+ " }\n",
753
+ "</style>\n",
754
+ "<table border=\"1\" class=\"dataframe\">\n",
755
+ " <thead>\n",
756
+ " <tr style=\"text-align: right;\">\n",
757
+ " <th></th>\n",
758
+ " <th>text</th>\n",
759
+ " <th>label</th>\n",
760
+ " <th>language</th>\n",
761
+ " </tr>\n",
762
+ " </thead>\n",
763
+ " <tbody>\n",
764
+ " <tr>\n",
765
+ " <th>[2019-06-13 14:12:47]</th>\n",
766
+ " <td>t as vu ce que t as fait a la soiré sale pute</td>\n",
767
+ " <td>1</td>\n",
768
+ " <td>fr</td>\n",
769
+ " </tr>\n",
770
+ " <tr>\n",
771
+ " <th>[2019-06-13 14:14:56]</th>\n",
772
+ " <td>je suis espagnol pardon de ne pas tout savoir...</td>\n",
773
+ " <td>1</td>\n",
774
+ " <td>fr</td>\n",
775
+ " </tr>\n",
776
+ " <tr>\n",
777
+ " <th>[2019-06-13 14:15:18]</th>\n",
778
+ " <td>tu as embrasser Lucas alors que sophia est ta...</td>\n",
779
+ " <td>1</td>\n",
780
+ " <td>fr</td>\n",
781
+ " </tr>\n",
782
+ " <tr>\n",
783
+ " <th>[2019-06-13 14:15:27]</th>\n",
784
+ " <td>tu as foutu la merde dans leurs couple</td>\n",
785
+ " <td>1</td>\n",
786
+ " <td>fr</td>\n",
787
+ " </tr>\n",
788
+ " <tr>\n",
789
+ " <th>[2019-06-13 14:15:31]</th>\n",
790
+ " <td>c'est dégeulasse</td>\n",
791
+ " <td>1</td>\n",
792
+ " <td>fr</td>\n",
793
+ " </tr>\n",
794
+ " <tr>\n",
795
+ " <th>...</th>\n",
796
+ " <td>...</td>\n",
797
+ " <td>...</td>\n",
798
+ " <td>...</td>\n",
799
+ " </tr>\n",
800
+ " <tr>\n",
801
+ " <th>[2019-06-13 14:32:11]</th>\n",
802
+ " <td>salope</td>\n",
803
+ " <td>0</td>\n",
804
+ " <td>fr</td>\n",
805
+ " </tr>\n",
806
+ " <tr>\n",
807
+ " <th>[2019-06-13 14:16:49]</th>\n",
808
+ " <td>toi ta mere c satan</td>\n",
809
+ " <td>0</td>\n",
810
+ " <td>fr</td>\n",
811
+ " </tr>\n",
812
+ " <tr>\n",
813
+ " <th>[2019-06-13 14:22:15]</th>\n",
814
+ " <td>t conne</td>\n",
815
+ " <td>0</td>\n",
816
+ " <td>fr</td>\n",
817
+ " </tr>\n",
818
+ " <tr>\n",
819
+ " <th>[2019-06-13 14:22:22]</th>\n",
820
+ " <td>salope</td>\n",
821
+ " <td>0</td>\n",
822
+ " <td>fr</td>\n",
823
+ " </tr>\n",
824
+ " <tr>\n",
825
+ " <th>[2019-06-13 14:32:11]</th>\n",
826
+ " <td>salope</td>\n",
827
+ " <td>0</td>\n",
828
+ " <td>fr</td>\n",
829
+ " </tr>\n",
830
+ " </tbody>\n",
831
+ "</table>\n",
832
+ "<p>42660 rows × 3 columns</p>\n",
833
+ "</div>"
834
+ ],
835
+ "text/plain": [
836
+ " text \\\n",
837
+ "[2019-06-13 14:12:47] t as vu ce que t as fait a la soiré sale pute \n",
838
+ "[2019-06-13 14:14:56] je suis espagnol pardon de ne pas tout savoir... \n",
839
+ "[2019-06-13 14:15:18] tu as embrasser Lucas alors que sophia est ta... \n",
840
+ "[2019-06-13 14:15:27] tu as foutu la merde dans leurs couple \n",
841
+ "[2019-06-13 14:15:31] c'est dégeulasse \n",
842
+ "... ... \n",
843
+ "[2019-06-13 14:32:11] salope \n",
844
+ "[2019-06-13 14:16:49] toi ta mere c satan \n",
845
+ "[2019-06-13 14:22:15] t conne \n",
846
+ "[2019-06-13 14:22:22] salope \n",
847
+ "[2019-06-13 14:32:11] salope \n",
848
+ "\n",
849
+ " label language \n",
850
+ "[2019-06-13 14:12:47] 1 fr \n",
851
+ "[2019-06-13 14:14:56] 1 fr \n",
852
+ "[2019-06-13 14:15:18] 1 fr \n",
853
+ "[2019-06-13 14:15:27] 1 fr \n",
854
+ "[2019-06-13 14:15:31] 1 fr \n",
855
+ "... ... ... \n",
856
+ "[2019-06-13 14:32:11] 0 fr \n",
857
+ "[2019-06-13 14:16:49] 0 fr \n",
858
+ "[2019-06-13 14:22:15] 0 fr \n",
859
+ "[2019-06-13 14:22:22] 0 fr \n",
860
+ "[2019-06-13 14:32:11] 0 fr \n",
861
+ "\n",
862
+ "[42660 rows x 3 columns]"
863
+ ]
864
+ },
865
+ "execution_count": 47,
866
+ "metadata": {},
867
+ "output_type": "execute_result"
868
+ }
869
+ ],
870
+ "source": [
871
+ "ids = [\"5823\", \"6783\", \"8898\", \"1966\"]\n",
872
+ "\n",
873
+ "manual_clean_df = clean_df.loc[clean_df.index.isin(ids) == False]\n",
874
+ "manual_clean_df"
875
+ ]
876
+ },
877
+ {
878
+ "cell_type": "code",
879
+ "execution_count": 63,
880
+ "id": "fb79ecfb",
881
+ "metadata": {},
882
+ "outputs": [
883
+ {
884
+ "data": {
885
+ "text/html": [
886
+ "<div>\n",
887
+ "<style scoped>\n",
888
+ " .dataframe tbody tr th:only-of-type {\n",
889
+ " vertical-align: middle;\n",
890
+ " }\n",
891
+ "\n",
892
+ " .dataframe tbody tr th {\n",
893
+ " vertical-align: top;\n",
894
+ " }\n",
895
+ "\n",
896
+ " .dataframe thead th {\n",
897
+ " text-align: right;\n",
898
+ " }\n",
899
+ "</style>\n",
900
+ "<table border=\"1\" class=\"dataframe\">\n",
901
+ " <thead>\n",
902
+ " <tr style=\"text-align: right;\">\n",
903
+ " <th></th>\n",
904
+ " <th>text</th>\n",
905
+ " <th>label</th>\n",
906
+ " <th>language</th>\n",
907
+ " </tr>\n",
908
+ " </thead>\n",
909
+ " <tbody>\n",
910
+ " <tr>\n",
911
+ " <th>[2019-06-13 14:21:43]</th>\n",
912
+ " <td>apprend a écrire connasse</td>\n",
913
+ " <td>0</td>\n",
914
+ " <td>fr</td>\n",
915
+ " </tr>\n",
916
+ " <tr>\n",
917
+ " <th>[2019-06-13 14:21:42]</th>\n",
918
+ " <td>il leur paie des apparts privés</td>\n",
919
+ " <td>0</td>\n",
920
+ " <td>fr</td>\n",
921
+ " </tr>\n",
922
+ " <tr>\n",
923
+ " <th>3333</th>\n",
924
+ " <td>la pauvreté n'est le collateral de la violence...</td>\n",
925
+ " <td>0</td>\n",
926
+ " <td>fr</td>\n",
927
+ " </tr>\n",
928
+ " <tr>\n",
929
+ " <th>827</th>\n",
930
+ " <td>\"Les décès massifs qui ont lieu en Afghanistan...</td>\n",
931
+ " <td>0</td>\n",
932
+ " <td>fr</td>\n",
933
+ " </tr>\n",
934
+ " <tr>\n",
935
+ " <th>2674</th>\n",
936
+ " <td>@Arashi_Storm @Jubaieuh par contre dire que je...</td>\n",
937
+ " <td>0</td>\n",
938
+ " <td>fr</td>\n",
939
+ " </tr>\n",
940
+ " <tr>\n",
941
+ " <th>1023</th>\n",
942
+ " <td>@TeTr4_Live Non mais totalement... Bon Zemmour...</td>\n",
943
+ " <td>0</td>\n",
944
+ " <td>fr</td>\n",
945
+ " </tr>\n",
946
+ " <tr>\n",
947
+ " <th>284</th>\n",
948
+ " <td>homme si une dame a essayé de se mettre entre ...</td>\n",
949
+ " <td>0</td>\n",
950
+ " <td>fr</td>\n",
951
+ " </tr>\n",
952
+ " <tr>\n",
953
+ " <th>664</th>\n",
954
+ " <td>rt l'histoire du quebec vaut la peine qu'on s...</td>\n",
955
+ " <td>0</td>\n",
956
+ " <td>fr</td>\n",
957
+ " </tr>\n",
958
+ " <tr>\n",
959
+ " <th>[2019-06-13 14:28:44]</th>\n",
960
+ " <td>déso j'avais plus de connexion</td>\n",
961
+ " <td>0</td>\n",
962
+ " <td>fr</td>\n",
963
+ " </tr>\n",
964
+ " <tr>\n",
965
+ " <th>609</th>\n",
966
+ " <td>@CommanderTorenn On va faire court,je ne m,int...</td>\n",
967
+ " <td>0</td>\n",
968
+ " <td>fr</td>\n",
969
+ " </tr>\n",
970
+ " <tr>\n",
971
+ " <th>[2019-06-13 14:35:20]</th>\n",
972
+ " <td>au moins j'arrive à me faire embrasse Diana</td>\n",
973
+ " <td>0</td>\n",
974
+ " <td>fr</td>\n",
975
+ " </tr>\n",
976
+ " <tr>\n",
977
+ " <th>[2019-06-13 14:32:35]</th>\n",
978
+ " <td>Mais oui...</td>\n",
979
+ " <td>0</td>\n",
980
+ " <td>fr</td>\n",
981
+ " </tr>\n",
982
+ " <tr>\n",
983
+ " <th>137</th>\n",
984
+ " <td>@djoulye @LiorahDouge @barriere_dr @MahmoudZur...</td>\n",
985
+ " <td>0</td>\n",
986
+ " <td>fr</td>\n",
987
+ " </tr>\n",
988
+ " <tr>\n",
989
+ " <th>1293</th>\n",
990
+ " <td>ptdrrr tout les «anti satanistes/theorie du co...</td>\n",
991
+ " <td>0</td>\n",
992
+ " <td>fr</td>\n",
993
+ " </tr>\n",
994
+ " <tr>\n",
995
+ " <th>[2019-06-13 14:35:04]</th>\n",
996
+ " <td>au moins mon pere est pas mort</td>\n",
997
+ " <td>0</td>\n",
998
+ " <td>fr</td>\n",
999
+ " </tr>\n",
1000
+ " <tr>\n",
1001
+ " <th>1602</th>\n",
1002
+ " <td>@LamoloQc @tvanouvelles On ne sera plus là pou...</td>\n",
1003
+ " <td>0</td>\n",
1004
+ " <td>fr</td>\n",
1005
+ " </tr>\n",
1006
+ " <tr>\n",
1007
+ " <th>[2019-06-13 14:16:49]</th>\n",
1008
+ " <td>toi ta mere c satan</td>\n",
1009
+ " <td>1</td>\n",
1010
+ " <td>fr</td>\n",
1011
+ " </tr>\n",
1012
+ " <tr>\n",
1013
+ " <th>[2019-06-13 14:24:51]</th>\n",
1014
+ " <td>t'es venu au nom de dieu Herbert?</td>\n",
1015
+ " <td>0</td>\n",
1016
+ " <td>fr</td>\n",
1017
+ " </tr>\n",
1018
+ " <tr>\n",
1019
+ " <th>[2019-06-13 14:11:46]</th>\n",
1020
+ " <td>Salut Adri, ça va?</td>\n",
1021
+ " <td>0</td>\n",
1022
+ " <td>fr</td>\n",
1023
+ " </tr>\n",
1024
+ " <tr>\n",
1025
+ " <th>[2019-06-13 14:32:11]</th>\n",
1026
+ " <td>salope</td>\n",
1027
+ " <td>0</td>\n",
1028
+ " <td>fr</td>\n",
1029
+ " </tr>\n",
1030
+ " </tbody>\n",
1031
+ "</table>\n",
1032
+ "</div>"
1033
+ ],
1034
+ "text/plain": [
1035
+ " text \\\n",
1036
+ "[2019-06-13 14:21:43] apprend a écrire connasse \n",
1037
+ "[2019-06-13 14:21:42] il leur paie des apparts privés \n",
1038
+ "3333 la pauvreté n'est le collateral de la violence... \n",
1039
+ "827 \"Les décès massifs qui ont lieu en Afghanistan... \n",
1040
+ "2674 @Arashi_Storm @Jubaieuh par contre dire que je... \n",
1041
+ "1023 @TeTr4_Live Non mais totalement... Bon Zemmour... \n",
1042
+ "284 homme si une dame a essayé de se mettre entre ... \n",
1043
+ "664 rt l'histoire du quebec vaut la peine qu'on s... \n",
1044
+ "[2019-06-13 14:28:44] déso j'avais plus de connexion \n",
1045
+ "609 @CommanderTorenn On va faire court,je ne m,int... \n",
1046
+ "[2019-06-13 14:35:20] au moins j'arrive à me faire embrasse Diana \n",
1047
+ "[2019-06-13 14:32:35] Mais oui... \n",
1048
+ "137 @djoulye @LiorahDouge @barriere_dr @MahmoudZur... \n",
1049
+ "1293 ptdrrr tout les «anti satanistes/theorie du co... \n",
1050
+ "[2019-06-13 14:35:04] au moins mon pere est pas mort \n",
1051
+ "1602 @LamoloQc @tvanouvelles On ne sera plus là pou... \n",
1052
+ "[2019-06-13 14:16:49] toi ta mere c satan \n",
1053
+ "[2019-06-13 14:24:51] t'es venu au nom de dieu Herbert? \n",
1054
+ "[2019-06-13 14:11:46] Salut Adri, ça va? \n",
1055
+ "[2019-06-13 14:32:11] salope \n",
1056
+ "\n",
1057
+ " label language \n",
1058
+ "[2019-06-13 14:21:43] 0 fr \n",
1059
+ "[2019-06-13 14:21:42] 0 fr \n",
1060
+ "3333 0 fr \n",
1061
+ "827 0 fr \n",
1062
+ "2674 0 fr \n",
1063
+ "1023 0 fr \n",
1064
+ "284 0 fr \n",
1065
+ "664 0 fr \n",
1066
+ "[2019-06-13 14:28:44] 0 fr \n",
1067
+ "609 0 fr \n",
1068
+ "[2019-06-13 14:35:20] 0 fr \n",
1069
+ "[2019-06-13 14:32:35] 0 fr \n",
1070
+ "137 0 fr \n",
1071
+ "1293 0 fr \n",
1072
+ "[2019-06-13 14:35:04] 0 fr \n",
1073
+ "1602 0 fr \n",
1074
+ "[2019-06-13 14:16:49] 1 fr \n",
1075
+ "[2019-06-13 14:24:51] 0 fr \n",
1076
+ "[2019-06-13 14:11:46] 0 fr \n",
1077
+ "[2019-06-13 14:32:11] 0 fr "
1078
+ ]
1079
+ },
1080
+ "execution_count": 63,
1081
+ "metadata": {},
1082
+ "output_type": "execute_result"
1083
+ }
1084
+ ],
1085
+ "source": [
1086
+ "manual_clean_df.sample(20)"
1087
+ ]
1088
+ },
1089
+ {
1090
+ "cell_type": "code",
1091
+ "execution_count": 69,
1092
+ "id": "922e8862",
1093
+ "metadata": {},
1094
+ "outputs": [],
1095
+ "source": [
1096
+ "manual_clean_df.to_csv(output_path)"
1097
+ ]
1098
+ }
1099
+ ],
1100
+ "metadata": {
1101
+ "kernelspec": {
1102
+ "display_name": "sexism_detection",
1103
+ "language": "python",
1104
+ "name": "sexism_detection"
1105
+ },
1106
+ "language_info": {
1107
+ "codemirror_mode": {
1108
+ "name": "ipython",
1109
+ "version": 3
1110
+ },
1111
+ "file_extension": ".py",
1112
+ "mimetype": "text/x-python",
1113
+ "name": "python",
1114
+ "nbconvert_exporter": "python",
1115
+ "pygments_lexer": "ipython3",
1116
+ "version": "3.9.15"
1117
+ }
1118
+ },
1119
+ "nbformat": 4,
1120
+ "nbformat_minor": 5
1121
+ }