language_detector 0.1.0

data/lib/textcat_ngrams/serbian-ascii.lm

@@ -0,0 +1,400 @@
+_	 34122
+a	 9113
+o	 8135
+i	 7736
+e	 7535
+n	 5207
+s	 4860
+j	 3995
+t	 3797
+r	 3660
+u	 3224
+l	 3065
+d	 3061
+e_	 2941
+v	 2786
+a_	 2746
+k	 2701
+m	 2492
+o_	 2328
+y	 2238
+p	 2151
+_s	 2148
+i_	 2094
+je	 1887
+c	 1854
+z	 1731
+_n	 1437
+_p	 1432
+g	 1418
+b	 1368
+u_	 1333
+je_	 1237
+,	 1226
+,_	 1214
+_i	 1130
+st	 1105
+na	 1076
+.	 1056
+_d	 1033
+._	 1030
+_j	 967
+ra	 934
+ko	 908
+ni	 900
+cy	 893
+sy	 875
+_je	 871
+_o	 824
+ta	 799
+no	 780
+_u	 777
+re	 766
+_b	 764
+_k	 763
+da	 760
+ne	 754
+li	 750
+ti	 745
+se	 722
+po	 713
+to	 713
+_je_	 696
+an	 688
+ja	 683
+pr	 665
+va	 651
+lo	 634
+_z	 626
+m_	 625
+is	 625
+il	 622
+ov	 621
+la	 621
+_m	 615
+bi	 604
+_t	 603
+_po	 594
+en	 586
+_se	 578
+os	 578
+in	 576
+od	 576
+ka	 552
+ve	 548
+ij	 538
+_pr	 536
+al	 536
+vo	 535
+om	 530
+_i_	 525
+nj	 515
+ed	 509
+_na	 507
+na_	 503
+og	 499
+oj	 498
+ma	 493
+_bi	 492
+on	 489
+ak	 482
+im	 481
+ye	 481
+ro	 480
+vi	 473
+sa	 469
+ri	 464
+da_	 451
+av	 450
+at	 449
+se_	 447
+es	 446
+h	 443
+ao	 441
+ji	 437
+yi	 436
+_da	 433
+ad	 432
+_se_	 430
+lj	 428
+zy	 426
+za	 426
+_ne	 425
+de	 422
+tr	 417
+cj	 415
+_u_	 414
+_c	 412
+le	 402
+_v	 397
+ar	 390
+_g	 390
+ic	 384
+n_	 382
+ju	 379
+lo_	 377
+aj	 376
+_ko	 369
+ao_	 366
+ek	 361
+_da_	 359
+et	 356
+go	 354
+iz	 346
+_za	 345
+_r	 344
+or	 342
+mo	 341
+el	 340
+as	 339
+ik	 336
+te	 332
+_sa	 329
+d_	 323
+am	 320
+me	 318
+sto	 317
+di	 315
+ec	 311
+ol	 310
+a,_	 307
+a,	 307
+_ni	 302
+ya	 296
+do	 295
+yt	 294
+su	 292
+syt	 289
+li_	 288
+sta	 286
+ije	 284
+ko_	 277
+ti_	 277
+la_	 277
+ga	 276
+bil	 275
+no_	 274
+a.	 273
+nu	 272
+a._	 271
+ne_	 271
+om_	 268
+_cy	 266
+_na_	 263
+_bil	 263
+sv	 263
+ru	 259
+to_	 256
+_od	 253
+cyi	 253
+nje	 251
+it	 251
+pa	 250
+az	 248
+e,_	 245
+e,	 245
+ob	 244
+dn	 243
+ac	 242
+ost	 242
+k_	 240
+iv	 239
+io	 238
+_su	 238
+_iz	 237
+ilo	 235
+_sv	 234
+_ka	 233
+koj	 231
+mi	 229
+im_	 229
+ije_	 227
+g_	 226
+em	 223
+su_	 223
+ih	 223
+ji_	 221
+kr	 220
+ut	 220
+_koj	 220
+V	 218
+_st	 218
+ye_	 217
+_l	 214
+_V	 213
+ovo	 211
+j_	 210
+uc	 208
+ja_	 208
+h_	 207
+nij	 206
+sk	 206
+ot	 203
+io_	 203
+gl	 203
+_do	 201
+ok	 200
+ns	 199
+ilo_	 199
+er	 197
+ih_	 195
+pre	 193
+ci	 193
+og_	 193
+ki	 192
+sl	 191
+t_	 189
+ni_	 189
+_a	 189
+vr	 188
+ati	 187
+_su_	 186
+nije	 181
+pro	 181
+be	 180
+yn	 179
+cye	 178
+ju_	 178
+ku	 177
+isy	 177
+ta_	 174
+sye	 172
+_tr	 172
+O	 172
+jen	 172
+_to	 171
+pi	 168
+_pre	 168
+S	 168
+ima	 167
+nije_	 167
+_mo	 166
+eg	 166
+e._	 164
+za_	 164
+e.	 164
+_pro	 164
+gov	 163
+N	 162
+dr	 162
+ako	 162
+tv	 162
+_S	 160
+P	 159
+ma_	 159
+_on	 159
+sp	 158
+nst	 158
+anj	 158
+dj	 157
+oc	 157
+_sy	 156
+ev	 155
+ce	 155
+lik	 154
+_nij	 153
+_N	 152
+ist	 151
+_P	 151
+_nije	 151
+-	 151
+ba	 150
+jed	 150
+sti	 150
+ova	 149
+_is	 148
+id	 148
+ton	 148
+ke	 147
+pos	 147
+od_	 147
+osy	 146
+Vi	 146
+ila	 145
+ins	 145
+bo	 145
+_Vi	 145
+ir	 144
+_za_	 144
+oz	 144
+ecj	 144
+cje	 143
+on_	 143
+zn	 142
+_O	 141
+us	 141
+i,	 141
+i,_	 141
+mu	 140
+inst	 140
+cya	 140
+oji	 139
+esy	 139
+icy	 139
+lja	 138
+_go	 138
+i.	 138
+_re	 137
+_bilo	 137
+edn	 137
+acy	 137
+rat	 137
+bilo	 137
+ali	 136
+ecy	 136
+ija	 135
+pri	 135
+ad_	 135
+lic	 135
+i._	 135
+Vins	 134
+Vin	 134
+ston	 134
+Vinst	 134
+ga_	 134
+nston	 134
+insto	 134
+nsto	 134
+_Vins	 133
+_Vin	 133
+zi	 132
+ran	 131
+le_	 130
+ili	 130
+bilo_	 130
+_pos	 129
+ila_	 129
+est	 128
+_ve	 128
+tre	 128
+zye	 127
+_nj	 127
+si	 126
+f	 126
+alo	 125
+ako_	 125
+tra	 125
+sa_	 125
+pu	 124
+ud	 124
+z_	 124
+_ra	 124
+iti	 124
+_de	 124
+odi	 123
+T	 123
+-_	 122
+o,_	 121
+o,	 121
+du	 121
+rs	 121
+B	 120
+ka_	 119
+red	 119
+_od_	 118
+an_	 118
+nu_	 118
+iko	 117
+dno	 117
+_pa	 117
+s_	 116

data/lib/textcat_ngrams/slovak-ascii.lm

@@ -0,0 +1,400 @@
+_	 20064
+a	 4991
+o	 4983
+e	 3838
+n	 3342
+i	 3317
+r	 2583
+s	 2501
+v	 2383
+t	 2325
+c	 1918
+k	 1912
+l	 1888
+d	 1736
+u	 1725
+p	 1543
+a_	 1527
+y	 1371
+m	 1339
+z	 1227
+h	 1194
+e_	 956
+_p	 881
+_s	 828
+o_	 814
+na	 809
+b	 808
+_v	 798
+j	 797
+.	 796
+ov	 795
+._	 785
+st	 687
+i_	 665
+,	 657
+,_	 657
+_n	 625
+ch	 623
+u_	 618
+ro	 617
+po	 598
+_o	 546
+ne	 529
+en	 520
+v_	 519
+_a	 516
+ra	 506
+pr	 488
+y_	 481
+od	 476
+_z	 467
+ie	 462
+ni	 448
+an	 447
+vy	 434
+to	 433
+h_	 431
+_na	 429
+re	 416
+ch_	 404
+ho	 401
+al	 399
+ci	 394
+va	 387
+na_	 386
+_pr	 385
+_d	 384
+_k	 378
+la	 377
+ko	 375
+do	 374
+_po	 373
+si	 353
+_t	 346
+_r	 337
+os	 336
+no	 334
+in	 328
+tr	 324
+om	 321
+_v_	 320
+ny	 319
+m_	 319
+ri	 317
+S	 306
+ac	 302
+sa	 300
+ti	 300
+_m	 298
+za	 298
+er	 291
+ia	 290
+ce	 290
+li	 289
+yc	 286
+ych	 285
+ed	 284
+at	 281
+ob	 281
+ak	 280
+_na_	 280
+il	 279
+_,	 279
+_,_	 279
+ok	 277
+sk	 268
+ych_	 265
+_c	 262
+mi	 261
+ol	 260
+me	 260
+l_	 259
+t_	 259
+ku	 258
+ta	 256
+le	 256
+_b	 254
+or	 252
+_a_	 250
+lo	 247
+oc	 246
+vo	 246
+es	 244
+ve	 242
+_vy	 240
+on	 238
+_sa	 231
+as	 231
+da	 230
+aj	 228
+av	 218
+el	 216
+ova	 216
+ic	 215
+ne_	 209
+_do	 208
+sa_	 207
+ka	 205
+_sa_	 204
+te	 203
+j_	 201
+_ro	 199
+P	 198
+_za	 196
+ky	 196
+_S	 195
+je	 194
+ar	 193
+_.	 193
+it	 192
+s_	 192
+em	 191
+ej	 191
+ur	 190
+ad	 189
+_o_	 187
+_._	 187
+ov_	 185
+de	 180
+_%	 179
+om_	 179
+%	 179
+_u	 174
+pre	 173
+dn	 173
+D	 172
+rok	 170
+ie_	 170
+sp	 169
+pri	 167
+_pre	 167
+am	 165
+-	 164
+ke	 164
+eh	 162
+oz	 161
+k_	 160
+ost	 160
+_j	 156
+f	 155
+zn	 153
+g	 152
+kt	 152
+ho_	 151
+eho	 151
+hod	 150
+ku_	 148
+is	 148
+zi	 147
+cn	 147
+eho_	 146
+ej_	 145
+az	 145
+tu	 145
+_pri	 144
+cen	 144
+_st	 143
+ma	 142
+ast	 141
+_ce	 140
+rov	 140
+la_	 138
+ot	 138
+nych	 135
+nyc	 135
+_ob	 133
+z_	 133
+nych_	 131
+N	 130
+li_	 129
+ani	 129
+co	 128
+nt	 128
+ny_	 127
+E	 125
+_ne	 124
+)	 123
+_(	 123
+sti	 123
+A	 123
+(	 123
+cho	 122
+vi	 122
+_sp	 122
+di	 120
+pa	 120
+n_	 119
+ju	 118
+ys	 117
+bo	 117
+_P	 116
+_tr	 115
+V	 114
+je_	 114
+ln	 114
+_i	 113
+ze	 113
+spo	 112
+_N	 112
+nd	 111
+nu	 111
+so	 111
+red	 110
+vn	 110
+kl	 110
+kov	 110
+_cen	 110
+_rok	 109
+tn	 109
+du	 109
+nc	 109
+ap	 109
+d_	 108
+van	 108
+ca	 108
+M	 108
+chod	 107
+ti_	 107
+U	 106
+_ak	 106
+ru	 105
+sta	 105
+ym	 105
+_-	 104
+et	 103
+_h	 102
+est	 102
+_je	 102
+nos	 101
+aci	 101
+us	 100
+dov	 100
+pod	 100
+_to	 100
+tor	 99
+uc	 99
+ras	 98
+ky_	 98
+_s_	 98
+_mi	 97
+*	 97
+uj	 97
+nost	 97
+vys	 97
+ovy	 97
+ez	 97
+oku	 96
+_V	 96
+op	 96
+bc	 96
+rast	 96
+se	 95
+B	 95
+roku	 95
+kto	 94
+ove	 94
+by	 94
+-_	 94
+_ko	 93
+obc	 92
+nie	 91
+ia_	 91
+ka_	 91
+*_	 90
+ali	 90
+lo_	 89
+ovan	 89
+to_	 88
+iz	 88
+_bo	 88
+_l	 88
+odo	 87
+bch	 87
+bchod	 87
+bcho	 87
+sl	 86
+st_	 86
+pred	 86
+C	 86
+pol	 85
+_pred	 85
+R	 85
+ik	 84
+uro	 84
+pi	 84
+ek	 84
+zo	 83
+eni	 83
+obch	 83
+cie	 83
+oku_	 83
+obcho	 83
+ns	 83
+roku_	 82
+ii	 82
+tv	 82
+ba	 82
+ent	 82
+_spo	 81
+tov	 81
+pe	 81
+kon	 80
+kc	 80
+ec	 80
+kci	 80
+ck	 80
+x	 79
+osti	 79
+_Sk	 79
+mi_	 79
+_in	 79
+Sk	 79
+sia	 79
+br	 78
+rh	 78
+val	 78
+olo	 77
+_pod	 77
+%_	 77
+_%_	 77
+bu	 77
+_f	 77
+iv	 77
+_obc	 77
+_obch	 77
+eb	 76
+str	 76
+nej	 76
+_D	 76
+ni_	 75
+ou	 75
+im	 75
+ena	 74
+tre	 74
+_A	 74
+mo	 74
+su	 74
+rz	 73
+_trh	 73
+trh	 73
+_U	 73
+al_	 73
+_ra	 73
+_e	 72
+_C	 72
+sti_	 72
+zv	 72
+te_	 72
+cno	 72
+oj	 72
+ktor	 71
+_roku	 71
+ocn	 71
+ina	 71
+sil	 71
+nov	 71
+alo	 71
+odn	 70
+nan	 70
+oh	 70