neelsahu commited on
Commit
65610be
1 Parent(s): a10a6f5

few words added into dictionary

Browse files
Files changed (1) hide show
  1. language_detection.py +48 -1
language_detection.py CHANGED
@@ -180,7 +180,54 @@ stop_words = ['i',
180
  'won',
181
  "won't",
182
  'wouldn',
183
- "wouldn't"]
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
184
  # Create a lemmatizer object
185
  lemmatizer = WordNetLemmatizer()
186
 
 
180
  'won',
181
  "won't",
182
  'wouldn',
183
+ "wouldn't",
184
+ "its",
185
+ "whats",
186
+ "im",
187
+ "youre",
188
+ "hes",
189
+ "shes",
190
+ "were",
191
+ "theyre",
192
+ "cant",
193
+ "dont",
194
+ "wont",
195
+ "isnt",
196
+ "arent",
197
+ "wasnt",
198
+ "werent",
199
+ "couldnt",
200
+ "shouldnt",
201
+ "wouldnt",
202
+ "ive",
203
+ "youve",
204
+ "weve",
205
+ "theyve",
206
+ "id",
207
+ "youd",
208
+ "lets",
209
+ "thats",
210
+ "theres",
211
+ "heres",
212
+ "ill",
213
+ "hell",
214
+ "shell",
215
+ "mustnt",
216
+ "mightnt",
217
+ "shant",
218
+ "neednt",
219
+ "oclock",
220
+ "cause",
221
+ "gimme",
222
+ "wanna",
223
+ "gonna",
224
+ "kinda",
225
+ "sorta",
226
+ "lemme",
227
+ "aint",
228
+ "dunno",
229
+ "gotta",
230
+ "yall"]
231
  # Create a lemmatizer object
232
  lemmatizer = WordNetLemmatizer()
233