language_detector 0.1.0

data/lib/textcat_ngrams/latvian.lm

@@ -0,0 +1,400 @@
+_	 24504
+a	 8735
+i	 7353
+s	 6064
+t	 5176
+e	 4390
+r	 4339
+u	 4200
+n	 3422
+�	 3055
+m	 2817
+p	 2755
+k	 2703
+o	 2643
+s_	 2595
+d	 2131
+l	 2035
+j	 1972
+�	 1958
+ie	 1887
+v	 1822
+as	 1530
+u_	 1396
+_p	 1380
+b	 1357
+z	 1311
+�	 1267
+a_	 1261
+as_	 1217
+.	 1065
+,	 1048
+,_	 1033
+g	 1010
+ar	 970
+�	 965
+ti	 962
+c	 922
+_a	 876
+ja	 838
+st	 828
+_i	 822
+_v	 805
+pa	 792
+_t	 775
+._	 774
+um	 760
+_k	 749
+�_	 734
+_n	 728
+es	 712
+i_	 706
+at	 703
+is	 695
+_s	 680
+ai	 649
+ta	 635
+an	 631
+ka	 621
+r_	 615
+�b	 586
+ij	 585
+_u	 568
+_pa	 548
+un	 541
+no	 539
+va	 537
+ma	 525
+ra	 522
+ri	 515
+iz	 512
+in	 509
+vi	 508
+pr	 506
+sa	 487
+ju	 486
+t�	 483
+ik	 473
+n_	 472
+am	 454
+en	 452
+tie	 440
+na	 432
+�a	 430
+tu	 420
+ija	 420
+ir	 419
+o_	 415
+m_	 414
+li	 411
+ci	 409
+ku	 404
+mi	 401
+_no	 398
+ek	 398
+_un	 397
+t_	 395
+_l	 394
+un_	 393
+_d	 393
+ot	 392
+_un_	 392
+ies	 391
+ar_	 388
+n�	 387
+da	 387
+ei	 385
+em	 382
+_ti	 382
+pi	 380
+al	 368
+ba	 356
+ts	 355
+_iz	 353
+ap	 350
+jas	 349
+ko	 346
+au	 345
+�an	 339
+re	 339
+ni	 337
+k�	 328
+ro	 327
+et	 326
+r�	 325
+ga	 323
+mu	 322
+jas_	 321
+tr	 317
+_va	 312
+te	 310
+_pr	 307
+sp	 298
+sk	 297
+ne	 295
+�ba	 293
+�s	 289
+ijas	 287
+_ar	 283
+la	 283
+to	 282
+ad	 280
+dz	 278
+�g	 277
+t�	 277
+�j	 277
+_sa	 277
+j�	 274
+_m	 273
+ties	 273
+l�	 272
+ijas_	 269
+_at	 269
+ed	 267
+_pi	 267
+�	 267
+s,	 265
+s,_	 265
+�r	 261
+gu	 258
+f	 258
+si	 257
+m�	 256
+nt	 254
+tu_	 254
+_r	 253
+ru	 252
+�j	 251
+jum	 251
+�s_	 250
+�t	 249
+iem	 248
+_b	 248
+_vi	 245
+par	 244
+pie	 244
+rt	 243
+v�	 242
+_ne	 240
+�r	 240
+ai_	 239
+_.	 239
+_j	 237
+us	 237
+_ka	 236
+er	 232
+bas	 232
+_tie	 232
+_par	 230
+di	 229
+�	 228
+ms	 228
+�d	 226
+�bas	 224
+om	 220
+bu	 219
+umu	 217
+iek	 216
+kt	 216
+it	 216
+r�	 215
+_ko	 214
+ana	 211
+�k	 211
+or	 210
+de	 210
+in�	 207
+d�	 204
+ur	 203
+pri	 203
+_c	 202
+ta_	 202
+bas_	 201
+es_	 201
+_pie	 200
+il	 200
+os	 199
+�ana	 199
+id	 198
+L	 198
+_ie	 197
+bi	 194
+ak	 194
+ja_	 194
+�bas_	 193
+�t	 192
+�bu	 192
+_g	 191
+�	 189
+ma_	 189
+ien	 188
+kum	 187
+�	 187
+_ir	 186
+on	 186
+_ties	 185
+z�	 185
+ce	 185
+ied	 184
+ist	 183
+ts_	 183
+_�	 183
+�_	 182
+�m	 181
+vie	 180
+_�_	 180
+cij	 180
+ka_	 179
+_ap	 175
+s�	 174
+ir_	 174
+zi	 174
+uma	 173
+�l	 172
+_ir_	 172
+o�	 171
+_da	 169
+�_	 168
+ar�	 168
+)	 167
+do	 166
+ve	 166
+(	 165
+ls	 165
+bu_	 164
+val	 164
+mu_	 164
+isk	 163
+uz	 161
+av	 160
+par_	 160
+_v�	 159
+_par_	 158
+v�r	 158
+_ar_	 157
+rie	 157
+_uz	 157
+aj	 156
+�bu_	 156
+oj	 155
+k�_	 155
+ld	 154
+iet	 154
+iku	 154
+ks	 153
+du	 153
+ep	 153
+ms_	 152
+ec	 152
+V	 151
+ais	 150
+str	 150
+nie	 149
+am_	 149
+ums	 148
+_(	 147
+ju_	 146
+z_	 142
+ru_	 142
+�j	 142
+gum	 142
+u,_	 141
+u,	 141
+i�	 141
+uma_	 141
+�rt	 141
+�gu	 141
+me	 140
+d�	 140
+�t	 140
+ent	 139
+ikum	 138
+p�	 138
+em_	 138
+_L	 138
+lie	 136
+sta	 136
+r�_	 136
+lst	 136
+eik	 135
+se	 135
+s.	 134
+pro	 134
+rs	 134
+s._	 134
+tik	 134
+l�	 134
+ska	 133
+pap	 133
+kas	 133
+rm	 133
+�ju	 133
+p�	 132
+nu	 132
+T	 132
+p�	 131
+tei	 131
+nas	 131
+_V	 131
+jo	 131
+l�g	 131
+ut	 131
+iem_	 130
+ras	 129
+pu	 129
+_li	 129
+_ga	 128
+dar	 128
+_k�	 128
+umu_	 127
+�v	 126
+�gum	 126
+kas_	 126
+t�b	 125
+vai	 125
+l�gu	 125
+P	 125
+ar�_	 125
+ev	 124
+lai	 124
+�r	 124
+l�gum	 124
+tv	 123
+a�	 123
+_ar�_	 122
+_ar�	 122
+_vie	 121
+S	 121
+_T	 120
+�ju	 120
+teik	 120
+cija	 119
+rb	 119
+_�	 119
+j�_	 119
+nas_	 118
+t�j	 118
+v�rt	 118
+iec	 118
+_la	 117
+�d	 116
+_vai	 116
+�jum	 116
+vien	 116
+als	 116
+_l�	 116
+ap�r	 115
+rts	 115
+isi	 115
+p�r	 115
+pap�	 115
+pap�r	 115
+ap�	 115
+tsp	 115
+m�	 114
+im	 114
+be	 114
+is_	 114
+alst	 114
+_ja	 114
+rtspa	 113

data/lib/textcat_ngrams/lithuanian.lm

@@ -0,0 +1,400 @@
+_	 23070
+i	 8196
+a	 7763
+s	 5179
+e	 3458
+o	 3425
+u	 3371
+t	 3344
+r	 3240
+n	 3070
+k	 3019
+l	 2060
+p	 1889
+s_	 1756
+m	 1680
+d	 1640
+v	 1590
+,	 1501
+,_	 1497
+�	 1400
+g	 1302
+j	 1200
+o_	 1138
+.	 1128
+ai	 1113
+�	 1075
+_p	 1047
+_k	 1035
+i_	 990
+._	 984
+as	 975
+b	 937
+au	 903
+is	 863
+_n	 815
+ka	 811
+si	 804
+_s	 799
+ia	 796
+�	 759
+ta	 735
+ie	 731
+ti	 728
+y	 721
+_t	 711
+in	 697
+_i	 686
+us	 675
+pa	 658
+ir	 640
+ar	 633
+�	 620
+_a	 617
+r_	 609
+ne	 583
+a_	 578
+ri	 569
+_v	 558
+_pa	 541
+al	 540
+ra	 531
+li	 522
+�	 509
+u_	 493
+vi	 490
+_ka	 489
+da	 487
+_ne	 482
+_j	 481
+en	 478
+�_	 445
+os	 444
+ki	 442
+�	 441
+an	 440
+ik	 434
+ma	 433
+as_	 430
+st	 420
+_d	 420
+ai_	 417
+s,_	 396
+am	 396
+is_	 396
+s,	 396
+_ir	 390
+jo	 387
+ir_	 385
+at	 384
+_ir_	 383
+i�	 381
+ni	 377
+ga	 376
+_b	 367
+na	 366
+e_	 365
+to	 352
+pr	 337
+_g	 332
+_m	 332
+�	 328
+�	 326
+io	 323
+ei	 323
+�_	 321
+su	 320
+�i	 320
+uo	 319
+la	 317
+er	 303
+va	 299
+vo	 298
+�i	 293
+es	 291
+-	 290
+�_	 286
+-_	 285
+_vi	 284
+_-	 283
+ko	 283
+_-_	 283
+iau	 281
+sa	 278
+s.	 278
+iu	 276
+et	 274
+nu	 265
+ja	 262
+�_	 262
+_ta	 262
+s._	 259
+el	 258
+�j	 258
+os_	 256
+im	 255
+it	 254
+_pr	 253
+no	 252
+av	 251
+ur	 251
+_�	 248
+mi	 246
+ve	 245
+ak	 245
+ku	 243
+ek	 241
+tu	 241
+_�	 241
+ad	 241
+�	 238
+_l	 238
+_i�	 236
+us_	 236
+_su	 232
+oj	 231
+au_	 231
+ba	 224
+nt	 218
+me	 214
+te	 210
+jo_	 208
+ro	 205
+�s	 204
+i�	 203
+tai	 203
+_�	 201
+l�	 201
+pas	 200
+ip	 198
+iai	 197
+di	 195
+usi	 194
+vo_	 193
+kai	 190
+i,	 189
+i,_	 189
+sk	 187
+_r	 186
+aus	 184
+ap	 183
+gi	 182
+mo	 182
+uk	 181
+t_	 180
+_da	 177
+be	 176
+ien	 176
+o,	 174
+il	 174
+o,_	 173
+ti_	 173
+re	 172
+_nu	 171
+�ia	 171
+_pas	 169
+pri	 169
+�_	 169
+ok	 168
+d�	 166
+�l	 162
+bu	 161
+�i	 160
+on	 160
+asi	 159
+id	 158
+ul	 157
+ji	 157
+aip	 156
+ia_	 155
+_sa	 154
+ot	 154
+gal	 153
+nk	 153
+om	 152
+p_	 151
+_pri	 151
+od	 151
+vis	 150
+_at	 150
+vie	 150
+uv	 150
+ab	 150
+i�_	 149
+ama	 149
+n�	 149
+A	 149
+sta	 148
+kr	 148
+_A	 145
+ais	 144
+rt	 143
+J	 142
+_J	 142
+�jo	 142
+_�_	 141
+K	 140
+_K	 139
+pi	 139
+avo	 139
+ng	 139
+mas	 138
+du	 138
+ug	 137
+_ga	 137
+ol	 136
+tik	 135
+ali	 135
+..	 135
+_vis	 135
+iek	 134
+ini	 133
+kad	 133
+a,	 133
+le	 133
+k�	 133
+T	 133
+t�	 132
+i.	 132
+_T	 131
+d�	 131
+a,_	 131
+go	 131
+ip_	 131
+aip_	 131
+r�	 129
+a�	 129
+d�i	 129
+ke	 128
+k_	 128
+d_	 127
+ad_	 127
+_ti	 127
+_ma	 126
+_va	 125
+i._	 124
+z	 123
+�k	 123
+iau_	 121
+m_	 120
+kar	 119
+um	 119
+t�	 119
+_kad	 118
+u�	 117
+tr	 117
+_kai	 117
+as,	 116
+og	 116
+kad_	 116
+as,_	 116
+_ji	 116
+rie	 115
+yt	 114
+_ja	 114
+_ko	 114
+v�	 114
+ys	 114
+jau	 113
+ar_	 113
+pra	 112
+a�	 112
+ant	 112
+kl	 111
+tas	 111
+gu	 111
+_ku	 110
+si_	 110
+pe	 109
+eb	 109
+N	 109
+�jo_	 108
+_kad_	 108
+se	 108
+_ki	 108
+or	 107
+ly	 107
+lo	 107
+iai_	 107
+_la	 106
+ts	 106
+�s	 106
+ig	 105
+_N	 105
+ut	 105
+_u	 105
+�s_	 105
+avo_	 104
+ie_	 104
+a.	 104
+kt	 104
+em	 103
+do	 103
+je	 102
+ks	 102
+ats	 101
+_be	 101
+ris	 101
+l_	 101
+_vie	 101
+tin	 101
+ag	 100
+dam	 100
+_st	 100
+a._	 99
+_bu	 99
+?	 99
+V	 98
+io_	 98
+kas	 98
+_jo	 98
+vien	 97
+_gal	 97
+�t	 97
+kia	 97
+lia	 96
+c	 96
+uri	 96
+_V	 96
+?_	 96
+uvo	 95
+ru	 95
+ty	 95
+ep	 94
+nd	 94
+lai	 94
+_tai	 94
+ju	 93
+man	 92
+o.	 92
+buv	 92
+t�	 92
+_�i	 91
+i�_	 91
+_i�_	 91
+i�k	 90
+o._	 90
+_to	 90
+bi	 90
+up	 89
+po	 89
+rs	 89
+_prie	 88
+�,_	 88
+aik	 88
+ui	 88
+_tik	 88
+ij	 88
+�,	 88
+pat	 88
+prie	 88
+I	 87
+ies	 87
+tai_	 87
+n_	 87
+pasi	 87
+ius	 87
+�t	 87
+eik	 86
+_me	 86
+ina	 86