Sentiment bias towards countries#

This tutorial is available as an IPython notebook at Malaya/example/sentiment-bias-towards-countries.

This module trained on both standard and local (included social media) language structures, so it is save to use for both.

[1]:

%%time
import malaya

CPU times: user 2.9 s, sys: 3.84 s, total: 6.74 s
Wall time: 1.97 s

/home/husein/dev/malaya/malaya/tokenizer.py:214: FutureWarning: Possible nested set at position 3397
  self.tok = re.compile(r'({})'.format('|'.join(pipeline)))
/home/husein/dev/malaya/malaya/tokenizer.py:214: FutureWarning: Possible nested set at position 3927
  self.tok = re.compile(r'({})'.format('|'.join(pipeline)))

This notebook simply want to test the bias of sentiment model given a text,

movie ni dirakam di <negara>.

[2]:

model = malaya.sentiment.huggingface()

Special tokens have been added in the vocabulary, make sure the associated word embeddings are fine-tuned or trained.

[4]:

model.predict_proba(['movie ni dirakam di Malaysia',
                    'movie ni dirakam di Israel'])

[4]:

[{'negative': 0.971319854259491,
  'neutral': 0.019404958933591843,
  'positive': 0.009275294840335846},
 {'negative': 0.8590973615646362,
  'neutral': 0.040735069662332535,
  'positive': 0.10016759485006332}]

[9]:

# !wget https://datahub.io/core/geo-countries/r/countries.geojson

[10]:

import json

with open('countries.geojson') as fopen:
    countries_json = json.load(fopen)

[11]:

from tqdm import tqdm

reviews = []
country_names = []
sentiments = []
for feature in tqdm(countries_json['features']):
    country_name = feature['properties']['ADMIN']
    country_names.append(country_name)
    text = f'movie ni dirakam di {country_name}'
    reviews.append(text)
    sentiments.append(model.predict_proba([text])[0]['positive'])

100%|████████████████████████████████████████| 255/255 [00:01<00:00, 140.87it/s]

[12]:

import pandas as pd
pd.set_option('display.max_rows', None)

[13]:

df = pd.DataFrame({'Country': country_names,
                   'Positive class probability': sentiments})
df

[13]:
Country Positive class probability
0 Aruba 0.039369
1 Afghanistan 0.169634
2 Angola 0.144256
3 Anguilla 0.356770
4 Albania 0.023863
5 Aland 0.044475
6 Andorra 0.130069
7 United Arab Emirates 0.050178
8 Argentina 0.227486
9 Armenia 0.029338
10 American Samoa 0.616004
11 Antarctica 0.009058
12 Ashmore and Cartier Islands 0.079729
13 French Southern and Antarctic Lands 0.027264
14 Antigua and Barbuda 0.105469
15 Australia 0.067188
16 Austria 0.616803
17 Azerbaijan 0.030310
18 Burundi 0.038826
19 Belgium 0.116550
20 Benin 0.073243
21 Burkina Faso 0.028318
22 Bangladesh 0.079000
23 Bulgaria 0.655491
24 Bahrain 0.036841
25 The Bahamas 0.012471
26 Bosnia and Herzegovina 0.106377
27 Bajo Nuevo Bank (Petrel Is.) 0.887322
28 Saint Barthelemy 0.153432
29 Belarus 0.038862
30 Belize 0.097082
31 Bermuda 0.047066
32 Bolivia 0.202820
33 Brazil 0.123829
34 Barbados 0.373711
35 Brunei 0.380603
36 Bhutan 0.974151
37 Botswana 0.064480
38 Central African Republic 0.412737
39 Canada 0.034126
40 Switzerland 0.013300
41 Chile 0.455905
42 China 0.038819
43 Ivory Coast 0.060523
44 Clipperton Island 0.941178
45 Cameroon 0.302088
46 Cyprus No Mans Area 0.005227
47 Democratic Republic of the Congo 0.989785
48 Republic of Congo 0.992472
49 Cook Islands 0.781634
50 Colombia 0.078325
51 Comoros 0.552591
52 Cape Verde 0.026867
53 Costa Rica 0.146759
54 Coral Sea Islands 0.143805
55 Cuba 0.034538
56 Curaçao 0.142810
57 Cayman Islands 0.140016
58 Northern Cyprus 0.078509
59 Cyprus 0.048920
60 Czech Republic 0.528698
61 Germany 0.039214
62 Djibouti 0.038966
63 Dominica 0.185570
64 Denmark 0.431805
65 Dominican Republic 0.201829
66 Algeria 0.034873
67 Ecuador 0.128374
68 Egypt 0.246791
69 Eritrea 0.516720
70 Dhekelia Sovereign Base Area 0.002587
71 Spain 0.069895
72 Estonia 0.003157
73 Ethiopia 0.049434
74 Finland 0.229206
75 Fiji 0.033329
76 Falkland Islands 0.030577
77 France 0.042942
78 Faroe Islands 0.098745
79 Federated States of Micronesia 0.000912
80 Gabon 0.463258
81 United Kingdom 0.009613
82 Georgia 0.028626
83 Guernsey 0.167616
84 Ghana 0.208999
85 Gibraltar 0.013775
86 Guinea 0.018618
87 Gambia 0.103650
88 Guinea Bissau 0.022831
89 Equatorial Guinea 0.014294
90 Greece 0.296132
91 Grenada 0.140380
92 Greenland 0.377565
93 Guatemala 0.039723
94 Guam 0.046184
95 Guyana 0.091745
96 Hong Kong S.A.R. 0.135367
97 Heard Island and McDonald Islands 0.747990
98 Honduras 0.102065
99 Croatia 0.055273
100 Haiti 0.042153
101 Hungary 0.667065
102 Indonesia 0.211004
103 Isle of Man 0.166882
104 India 0.089701
105 Indian Ocean Territories 0.012646
106 British Indian Ocean Territory 0.002079
107 Ireland 0.367753
108 Iran 0.036559
109 Iraq 0.168938
110 Iceland 0.039860
111 Israel 0.100167
112 Italy 0.046229
113 Jamaica 0.064513
114 Jersey 0.146864
115 Jordan 0.043022
116 Japan 0.042664
117 Baykonur Cosmodrome 0.060239
118 Siachen Glacier 0.778532
119 Kazakhstan 0.082745
120 Kenya 0.210720
121 Kyrgyzstan 0.386843
122 Cambodia 0.133686
123 Kiribati 0.095092
124 Saint Kitts and Nevis 0.615573
125 South Korea 0.859220
126 Kosovo 0.308428
127 Kuwait 0.040864
128 Laos 0.042060
129 Lebanon 0.096411
130 Liberia 0.240108
131 Libya 0.083740
132 Saint Lucia 0.506887
133 Liechtenstein 0.092944
134 Sri Lanka 0.083067
135 Lesotho 0.114256
136 Lithuania 0.118200
137 Luxembourg 0.030526
138 Latvia 0.166041
139 Macao S.A.R 0.054484
140 Saint Martin 0.480976
141 Morocco 0.014966
142 Monaco 0.090793
143 Moldova 0.038843
144 Madagascar 0.699353
145 Maldives 0.086702
146 Mexico 0.013596
147 Marshall Islands 0.145141
148 Macedonia 0.016141
149 Mali 0.028769
150 Malta 0.049830
151 Myanmar 0.168810
152 Montenegro 0.080645
153 Mongolia 0.050451
154 Northern Mariana Islands 0.083891
155 Mozambique 0.052984
156 Mauritania 0.047578
157 Montserrat 0.024121
158 Mauritius 0.066048
159 Malawi 0.208873
160 Malaysia 0.009275
161 Namibia 0.020265
162 New Caledonia 0.064238
163 Niger 0.020084
164 Norfolk Island 0.247524
165 Nigeria 0.067220
166 Nicaragua 0.053066
167 Niue 0.058268
168 Netherlands 0.029782
169 Norway 0.661565
170 Nepal 0.530866
171 Nauru 0.046705
172 New Zealand 0.063832
173 Oman 0.056455
174 Pakistan 0.088653
175 Panama 0.169475
176 Pitcairn Islands 0.282857
177 Peru 0.051750
178 Spratly Islands 0.026154
179 Philippines 0.087797
180 Palau 0.176909
181 Papua New Guinea 0.026789
182 Poland 0.062732
183 Puerto Rico 0.222873
184 North Korea 0.315357
185 Portugal 0.122907
186 Paraguay 0.141607
187 Palestine 0.265100
188 French Polynesia 0.061339
189 Qatar 0.097631
190 Romania 0.743275
191 Russia 0.006057
192 Rwanda 0.123763
193 Western Sahara 0.054943
194 Saudi Arabia 0.188411
195 Scarborough Reef 0.069580
196 Sudan 0.145442
197 South Sudan 0.182804
198 Senegal 0.041048
199 Serranilla Bank 0.404349
200 Singapore 0.266856
201 South Georgia and South Sandwich Islands 0.149831
202 Saint Helena 0.428387
203 Solomon Islands 0.344656
204 Sierra Leone 0.100867
205 El Salvador 0.176265
206 San Marino 0.038682
207 Somaliland 0.040974
208 Somalia 0.044995
209 Saint Pierre and Miquelon 0.699458
210 Republic of Serbia 0.042408
211 Sao Tome and Principe 0.915933
212 Suriname 0.041542
213 Slovakia 0.113862
214 Slovenia 0.753359
215 Sweden 0.001290
216 Swaziland 0.288331
217 Sint Maarten 0.081521
218 Seychelles 0.126393
219 Syria 0.028406
220 Turks and Caicos Islands 0.012284
221 Chad 0.067085
222 Togo 0.065655
223 Thailand 0.034735
224 Tajikistan 0.023857
225 Turkmenistan 0.006858
226 East Timor 0.038230
227 Tonga 0.345105
228 Trinidad and Tobago 0.020336
229 Tunisia 0.174806
230 Turkey 0.047196
231 Tuvalu 0.041594
232 Taiwan 0.121632
233 United Republic of Tanzania 0.922501
234 Uganda 0.066297
235 Ukraine 0.088036
236 United States Minor Outlying Islands 0.061749
237 Uruguay 0.025120
238 United States of America 0.098839
239 US Naval Base Guantanamo Bay 0.013662
240 Uzbekistan 0.090956
241 Vatican 0.024224
242 Saint Vincent and the Grenadines 0.798123
243 Venezuela 0.164420
244 British Virgin Islands 0.070114
245 United States Virgin Islands 0.022017
246 Vietnam 0.051927
247 Vanuatu 0.071065
248 Wallis and Futuna 0.026419
249 Akrotiri Sovereign Base Area 0.003673
250 Samoa 0.153828
251 Yemen 0.044515
252 South Africa 0.435813
253 Zambia 0.051639
254 Zimbabwe 0.044964 
[ ]: