language_detector 0.1.0

data/lib/textcat_ngrams/icelandic.lm

@@ -0,0 +1,400 @@
+_	 26104
+a	 6496
+r	 6044
+n	 5160
+i	 5123
+s	 3987
+e	 3891
+u	 3582
+t	 3300
+�	 3126
+l	 3071
+g	 2726
+m	 2459
+k	 2256
+f	 2230
+r_	 1967
+v	 1641
+ar	 1472
+�_	 1420
+_s	 1332
+�	 1248
+o	 1221
+�	 1197
+a_	 1155
+in	 1150
+i_	 1114
+h	 1023
+j	 975
+d	 974
+st	 929
+a�	 925
+.	 922
+n_	 875
+._	 875
+_v	 874
+m_	 854
+nn	 842
+_f	 840
+ur	 824
+_a	 806
+�	 785
+_h	 782
+�	 779
+i�	 771
+er	 765
+um	 727
+g_	 716
+y	 711
+_e	 709
+�	 672
+b	 672
+ir	 671
+ri	 670
+an	 667
+�	 650
+_�	 648
+a�_	 645
+u_	 633
+na	 631
+�_	 625
+ar_	 622
+_�	 608
+�_	 606
+_�	 595
+ta	 567
+ei	 566
+la	 558
+_m	 549
+_�_	 549
+um_	 547
+t_	 535
+ti	 529
+_o	 524
+ur_	 523
+_a�	 519
+i�_	 512
+�i	 496
+ve	 494
+og	 494
+�	 482
+og_	 479
+_og	 478
+_og_	 478
+nd	 469
+p	 464
+ra	 455
+un	 454
+ir_	 452
+_a�_	 451
+ni	 439
+en	 439
+�u	 439
+_�_	 436
+ng	 434
+il	 404
+ga	 395
+_t	 395
+nu	 393
+ki	 392
+ja	 383
+inn	 379
+_b	 375
+sk	 352
+s_	 350
+vi	 349
+r�	 347
+�a	 345
+ef	 339
+ag	 336
+_u	 330
+se	 324
+lu	 324
+af	 321
+_ve	 320
+tu	 318
+em	 307
+eg	 304
+nn_	 303
+_l	 303
+va	 301
+_k	 296
+,	 295
+ns	 292
+re	 292
+tt	 291
+,_	 291
+l_	 286
+am	 286
+es	 285
+yr	 285
+al	 281
+da	 277
+S	 275
+gu	 273
+_se	 271
+ver	 268
+_g	 266
+ing	 266
+_n	 262
+is	 258
+_er	 257
+sa	 256
+�ur	 255
+le	 255
+_st	 255
+_S	 254
+sem	 254
+ll	 254
+me	 253
+ha	 251
+li	 249
+kk	 249
+rs	 247
+_vi	 247
+rn	 246
+sl	 244
+gi	 243
+ss	 242
+rf	 241
+fy	 240
+�i_	 240
+mi	 238
+ka	 237
+ma	 231
+ld	 230
+�	 229
+rir	 227
+sta	 227
+fyr	 227
+v�	 227
+di	 226
+ru	 224
+var	 224
+_fy	 222
+ku	 221
+em_	 221
+nar	 220
+_sem_	 220
+_sem	 220
+sem_	 220
+he	 219
+yri	 217
+_fyr	 216
+si	 216
+yrir	 215
+au	 212
+er_	 212
+ek	 211
+_ha	 210
+�e	 209
+fyri	 207
+fyrir	 207
+_�e	 205
+fi	 204
+fr	 203
+ge	 201
+or	 200
+ne	 200
+ann	 198
+j�	 198
+_va	 196
+_fyri	 196
+_ver	 194
+fl	 192
+_er_	 191
+_um	 189
+�k	 188
+til	 187
+_he	 186
+fa	 186
+il_	 182
+_ti	 178
+_til	 177
+gar	 176
+_var	 176
+na_	 176
+�	 175
+e�	 171
+fu	 170
+nni	 169
+_me	 168
+ki_	 167
+vi�	 166
+ey	 165
+fn	 165
+arf	 164
+til_	 163
+st_	 162
+_til_	 162
+�a	 161
+num	 161
+_�a	 161
+as	 160
+_vi�	 160
+rt	 159
+el	 158
+u�	 156
+inn_	 155
+_um_	 154
+ra_	 153
+b�	 153
+tar	 151
+ta_	 151
+er�	 151
+�r	 148
+and	 148
+_sa	 146
+ig	 146
+_en	 146
+nga	 145
+rir_	 145
+us	 144
+jar	 143
+et	 143
+�r	 142
+_sk	 140
+ndi	 140
+��	 139
+var_	 139
+_r	 138
+av	 138
+�k	 137
+nna	 137
+�ur_	 136
+�l	 136
+ko	 135
+nin	 135
+vi�_	 135
+�l	 135
+ins	 134
+ik	 133
+E	 133
+K	 133
+yrir_	 133
+ns_	 133
+on	 133
+ein	 132
+_vi�_	 132
+�g	 132
+j�	 132
+�	 132
+�l	 132
+�s	 132
+_mi	 131
+f_	 131
+s�	 131
+sj	 131
+stu	 131
+nda	 130
+_var_	 130
+gr	 129
+�s	 128
+t�	 127
+ri_	 126
+haf	 126
+_s�	 125
+v�k	 124
+rin	 124
+te	 124
+r.	 124
+r._	 123
+H	 123
+nes	 123
+�t	 123
+ru_	 123
+kr	 122
+F	 122
+�r	 121
+num_	 121
+k_	 121
+a.	 121
+_H	 121
+_fr	 120
+_ge	 120
+r�	 120
+_E	 120
+_�	 120
+ug	 120
+ngu	 119
+an_	 119
+inga	 118
+_K	 118
+_haf	 118
+enn	 117
+ars	 117
+ri�	 117
+en_	 117
+sin	 116
+kur	 116
+it	 116
+�a_	 116
+ti_	 115
+r�i	 114
+tj	 114
+ni_	 114
+at	 114
+tarf	 114
+br	 113
+slu	 113
+kki	 113
+rg	 113
+m�	 113
+kv	 113
+_en_	 113
+a._	 112
+gar_	 112
+du	 112
+ju	 110
+ei�	 110
+und	 110
+lag	 110
+tur	 110
+ega	 109
+h�	 109
+��	 109
+gn	 109
+hef	 109
+kj	 109
+_hef	 109
+_sta	 108
+B	 108
+V	 108
+sam	 107
+_ei	 106
+_B	 106
+ft	 106
+ga_	 106
+G	 106
+_G	 105
+l�	 105
+kki_	 105
+star	 104
+in_	 104
+R	 104
+me�	 104
+_�r	 103
+_me�	 103
+ekk	 103
+inu	 103
+t�	 103
+_V	 103
+m.	 103
+a�i	 103
+j�	 102
+�j	 102
+_b�	 102
+ess	 102
+h�s	 101
+ut	 101
+gs	 101
+av�	 101
+mu	 101
+_R	 101
+_�	 101
+rst	 100
+�jar	 100
+leg	 100
+�ja	 100
+ja_	 99
+av�k	 99

data/lib/textcat_ngrams/indonesian.lm

@@ -0,0 +1,400 @@
+_	 19406
+a	 10666
+n	 5455
+e	 4535
+i	 4387
+r	 2936
+t	 2902
+an	 2853
+u	 2841
+k	 2761
+s	 2311
+m	 2178
+d	 2134
+g	 2105
+l	 1780
+a_	 1506
+n_	 1476
+ng	 1449
+p	 1397
+b	 1275
+an_	 1270
+o	 1246
+h	 1130
+i_	 1108
+er	 1038
+ka	 1032
+_d	 1006
+y	 997
+,	 951
+en	 941
+ar	 914
+,_	 900
+_m	 880
+ya	 842
+ta	 838
+ang	 797
+di	 787
+da	 773
+.	 754
+la	 742
+._	 738
+me	 732
+ak	 728
+_s	 718
+at	 690
+ra	 688
+ga	 683
+_k	 671
+_me	 650
+in	 628
+ah	 601
+_t	 583
+_p	 570
+g_	 551
+_b	 551
+_di	 548
+ng_	 544
+ma	 536
+se	 526
+tu	 511
+na	 506
+al	 500
+ri	 490
+as	 483
+k_	 482
+j	 480
+si	 470
+ny	 467
+h_	 457
+sa	 452
+ang_	 439
+it	 424
+kan	 423
+ti	 418
+_se	 417
+pe	 412
+S	 409
+ba	 407
+ke	 407
+em	 405
+men	 405
+be	 403
+un	 401
+te	 401
+am	 396
+pa	 395
+nya	 390
+_men	 374
+el	 374
+t_	 371
+_a	 360
+_i	 356
+u_	 355
+kan_	 345
+_ke	 339
+is	 335
+ah_	 324
+_S	 324
+eng	 321
+nga	 320
+ia	 318
+_pe	 316
+ha	 313
+ap	 311
+r_	 308
+w	 305
+li	 301
+_da	 300
+s_	 299
+P	 296
+nd	 290
+_be	 287
+ik	 283
+ja	 281
+yan	 281
+ad	 275
+ek	 273
+uk	 272
+di_	 270
+bu	 269
+ya_	 268
+yang	 268
+ak_	 266
+ber	 265
+_y	 265
+_ya	 264
+_P	 263
+ru	 260
+K	 259
+yang_	 256
+_yang	 256
+_yan	 256
+nt	 255
+de	 253
+_te	 250
+wa	 249
+et	 247
+at_	 246
+ara	 245
+gan	 243
+A	 237
+ari	 235
+ala	 230
+itu	 229
+c	 225
+ol	 225
+ni	 225
+us	 225
+dan	 224
+_K	 224
+M	 224
+B	 223
+ata	 222
+ai	 221
+ur	 219
+nya_	 217
+ua	 215
+_ka	 214
+_ber	 210
+eb	 209
+ran	 206
+D	 206
+ela	 206
+_di_	 205
+_l	 204
+ngan	 204
+ter	 203
+re	 201
+-	 199
+aka	 198
+l_	 194
+_A	 191
+era	 191
+a,	 191
+e_	 190
+ir	 187
+I	 186
+tan	 185
+_B	 184
+ut	 184
+ku	 183
+a,_	 183
+il	 182
+J	 181
+um	 180
+_it	 180
+_itu	 180
+_ta	 179
+su	 179
+dan_	 177
+es	 177
+on	 177
+or	 177
+_dan	 176
+lu	 174
+_M	 172
+tu_	 172
+_dan_	 172
+enga	 171
+mb	 169
+R	 169
+si_	 168
+per	 168
+gan_	 168
+ngan_	 165
+"	 162
+_ter	 162
+a.	 161
+man	 161
+gk	 160
+a._	 160
+asi	 160
+ngk	 160
+ep	 160
+ag	 159
+ul	 158
+da_	 157
+m_	 155
+du	 155
+ada	 153
+ki	 153
+rt	 150
+mp	 150
+T	 150
+ama	 148
+ing	 148
+na_	 147
+_J	 147
+_D	 145
+ung	 145
+ana	 145
+n,	 144
+ju	 144
+ud	 144
+rin	 143
+gi	 143
+aw	 141
+lah	 138
+lan	 138
+_sa	 136
+ri_	 136
+meng	 136
+_meng	 135
+_ma	 134
+n,_	 134
+awa	 134
+st	 134
+eka	 133
+mi	 133
+mu	 132
+_T	 132
+po	 131
+ge	 131
+ar_	 130
+id	 129
+ko	 129
+le	 128
+_h	 128
+ena	 127
+_j	 126
+emb	 126
+ina	 125
+_r	 124
+itu_	 124
+ay	 123
+ngg	 123
+gg	 123
+rang	 123
+pi	 120
+nan	 120
+_ba	 119
+_la	 119
+apa	 119
+_I	 118
+p_	 118
+bi	 117
+ai_	 117
+ta_	 116
+san	 116
+Ke	 116
+ro	 115
+eri	 114
+kar	 113
+lah_	 113
+_itu_	 112
+aya	 111
+i,	 110
+an,	 110
+ra_	 110
+_per	 110
+im	 110
+ika	 109
+isi	 109
+mem	 109
+tah	 108
+_Ke	 108
+ian	 108
+_mem	 108
+akan	 108
+Se	 108
+to	 107
+ab	 107
+ngka	 106
+rs	 106
+gka	 106
+uk_	 105
+seb	 104
+_de	 104
+pu	 104
+i,_	 104
+ita	 104
+nda	 103
+_ti	 103
+ni_	 103
+ca	 103
+_Se	 103
+ers	 103
+pen	 103
+ini	 102
+an,_	 102
+angk	 101
+uh	 101
+han	 101
+nta	 100
+_in	 99
+f	 99
+nj	 99
+ok	 99
+aga	 99
+_R	 98
+as_	 98
+tr	 98
+mer	 97
+lam	 97
+and	 97
+end	 96
+anga	 96
+ne	 96
+Sa	 96
+ka_	 96
+arin	 95
+gu	 95
+_ha	 94
+Z	 94
+al_	 94
+ga_	 94
+_Z	 93
+_pen	 93
+dar	 93
+Ad	 93
+i._	 93
+ada_	 93
+atan	 93
+tak	 93
+i.	 93
+ia_	 92
+asa	 92
+ap_	 92
+ari_	 92
+kat	 92
+_seb	 92
+_Za	 91
+Za	 91
+den	 91
+n.	 90
+_u	 90
+_Ad	 90
+engan	 89
+ib	 89
+any	 89
+n._	 89
+o_	 89
+Zar	 88
+Zari	 88
+rina	 88
+_Zar	 88
+_Zari	 88
+Zarin	 88
+arina	 88
+ks	 88
+angka	 87
+oli	 87
+eg	 87
+kt	 86
+_Sa	 86
+hu	 85
+ih	 85
+us_	 85
+adi	 85
+om	 85
+eba	 85
+anya	 85
+_bu	 84
+denga	 83
+L	 83
+ed	 83
+dak	 83
+deng	 83
+ma_	 82
+asi_	 82