language_detector 0.1.0

data/lib/textcat_ngrams/romanian.lm

@@ -0,0 +1,400 @@
+_	 20674
+a	 6376
+e	 5815
+i	 5746
+t	 3396
+r	 3280
+n	 3103
+u	 2835
+s	 2611
+c	 2582
+e_	 2235
+l	 2224
+o	 2149
+a_	 1974
+d	 1629
+m	 1528
+p	 1410
+i_	 1358
+in	 1308
+_c	 1167
+_s	 1118
+_d	 999
+re	 905
+ar	 898
+,	 791
+,_	 786
+_p	 785
+de	 771
+_a	 754
+te	 687
+_i	 667
+at	 654
+ti	 645
+ca	 639
+n_	 630
+ta	 617
+si	 614
+_de	 609
+f	 606
+st	 583
+ri	 581
+u_	 567
+nt	 553
+.	 542
+ra	 540
+_m	 534
+g	 528
+v	 516
+ul	 516
+de_	 513
+_in	 503
+b	 492
+_de_	 474
+._	 472
+le	 459
+l_	 444
+un	 443
+_si	 440
+es	 437
+tr	 426
+ea	 420
+t_	 412
+ce	 412
+ma	 407
+cu	 402
+er	 398
+_ca	 397
+si_	 388
+_f	 387
+_l	 383
+z	 382
+la	 381
+ne	 370
+sa	 364
+as	 360
+_e	 357
+in_	 356
+an	 352
+it	 351
+te_	 346
+or	 345
+el	 345
+ci	 339
+_si_	 333
+_n	 330
+are	 324
+pe	 319
+re_	 317
+al	 310
+_t	 309
+se	 304
+ic	 295
+ie	 290
+_u	 290
+ul_	 290
+ni	 289
+int	 285
+_o	 280
+en	 279
+ta_	 279
+ur	 261
+pa	 256
+co	 255
+_pe	 254
+ia	 252
+mi	 251
+pr	 249
+_ma	 249
+oa	 249
+me	 246
+lu	 246
+li	 241
+im	 238
+_in_	 237
+da	 237
+na	 237
+_sa	 235
+ac	 234
+-	 234
+ii	 232
+est	 231
+r_	 231
+h	 230
+_cu	 230
+le_	 229
+ai	 229
+ca_	 227
+il	 226
+ru	 223
+sc	 223
+_v	 221
+nu	 220
+tu	 220
+_un	 220
+nd	 220
+di	 219
+are_	 216
+to	 215
+am	 214
+on	 213
+o_	 208
+is	 208
+sa_	 203
+la_	 200
+_b	 200
+ste	 195
+et	 194
+ec	 191
+_r	 186
+car	 185
+ui	 180
+un_	 179
+lo	 178
+cu_	 177
+ei	 176
+e,	 175
+e,_	 175
+pe_	 171
+m_	 167
+_la	 166
+a,	 165
+a,_	 164
+_ce	 164
+rt	 163
+_co	 163
+ent	 162
+ro	 162
+ele	 162
+_pe_	 160
+po	 160
+ea_	 159
+"	 158
+ntr	 158
+_cu_	 158
+_pr	 157
+ut	 157
+nc	 156
+ata	 155
+care	 154
+um	 153
+au	 151
+va	 151
+_o_	 150
+_car	 150
+ii_	 145
+ind	 145
+_un_	 144
+os	 144
+ad	 141
+_la_	 140
+I	 140
+este	 138
+ste_	 138
+care_	 138
+ir	 137
+ga	 136
+ap	 136
+ol	 136
+ra_	 136
+_di	 134
+D	 134
+_care	 133
+se_	 133
+om	 133
+ara	 133
+ati	 133
+fi	 133
+_sa_	 131
+zi	 130
+vi	 130
+_ca_	 129
+_se	 128
+_nu	 128
+ai_	 127
+ch	 127
+pi	 124
+ve	 123
+fa	 122
+ot	 121
+_a_	 120
+este_	 120
+du	 119
+ine	 119
+s_	 118
+fo	 118
+_ci	 118
+ui_	 118
+ba	 118
+i,	 117
+i,_	 116
+ne_	 115
+us	 115
+_g	 115
+a.	 115
+fe	 114
+A	 114
+pu	 114
+ce_	 113
+ar_	 113
+_pa	 113
+oc	 112
+sta	 112
+lui	 112
+ns	 112
+em	 112
+'	 112
+oar	 112
+din	 111
+iu	 111
+_int	 111
+ate	 111
+mu	 111
+hi	 110
+ele_	 110
+mp	 109
+_D	 109
+S	 109
+sti	 108
+bi	 108
+ata_	 107
+ti_	 107
+tra	 107
+C	 107
+c_	 106
+tre	 106
+_al	 105
+rea	 105
+mai	 105
+j	 104
+a._	 104
+gi	 104
+e.	 103
+d_	 103
+_fa	 103
+E	 102
+mo	 102
+at_	 101
+_e_	 101
+nte	 101
+lt	 101
+sp	 101
+za	 100
+mai_	 100
+su	 99
+na_	 98
+tat	 97
+sin	 97
+ez	 96
+tru	 96
+e._	 96
+ie_	 96
+ia_	 96
+_re	 96
+tul	 96
+_fo	 96
+ina	 95
+art	 95
+_C	 95
+no	 95
+nu_	 94
+_es	 94
+_po	 94
+cr	 94
+inc	 93
+_da	 92
+_mai	 92
+lui_	 92
+_din	 92
+_est	 92
+pre	 91
+_mai_	 91
+io	 91
+chi	 91
+ge	 90
+pri	 90
+eu	 90
+uri	 90
+az	 90
+_nu_	 89
+_me	 89
+ct	 89
+au_	 88
+esc	 88
+ev	 88
+ei_	 88
+min	 87
+ace	 87
+op	 86
+ng	 86
+ici	 86
+_lu	 85
+ari	 85
+_mi	 84
+ita	 84
+_S	 84
+_tr	 84
+ere	 83
+or_	 83
+ast	 83
+ist	 83
+nt_	 83
+_se_	 82
+ou	 82
+tin	 82
+intr	 82
+con	 82
+do	 81
+_fi	 81
+str	 81
+am_	 80
+rat	 80
+ru_	 80
+ri_	 80
+par	 80
+oi	 80
+uc	 79
+ze	 79
+pl	 79
+res	 78
+_ac	 77
+ulu	 77
+din_	 76
+va_	 76
+ada	 76
+ului	 75
+_con	 75
+id	 75
+inte	 74
+ile	 73
+cit	 73
+_din_	 73
+lor	 73
+_"	 72
+ig	 72
+rin	 72
+da_	 72
+_st	 72
+-_	 71
+_-	 71
+it_	 71
+ani	 71
+nd_	 71
+ci_	 70
+ag	 70
+eri	 70
+i.	 70
+tru_	 70
+_ne	 70
+rm	 70
+P	 69
+_este	 69
+nta	 69
+bu	 69
+une	 69
+ma_	 69
+nti	 69
+imp	 68
+_-_	 68
+iv	 68
+ind_	 68

data/lib/textcat_ngrams/rumantsch.lm

@@ -0,0 +1,400 @@
+_	 10888
+a	 3490
+e	 2268
+i	 2196
+s	 2169
+n	 1961
+t	 1555
+r	 1510
+l	 1281
+u	 1249
+a_	 1155
+c	 1060
+d	 954
+o	 815
+s_	 744
+g	 726
+m	 686
+h	 620
+p	 617
+n_	 559
+v	 493
+_s	 480
+er	 476
+ch	 469
+_d	 457
+in	 455
+da	 418
+_c	 404
+r_	 388
+_e	 377
+.	 358
+_p	 340
+as	 333
+l_	 328
+,	 327
+._	 320
+_i	 320
+,_	 315
+la	 313
+en	 312
+sc	 310
+an	 301
+ta	 300
+_da	 285
+f	 282
+_t	 268
+_a	 262
+nt	 259
+_m	 252
+un	 248
+ra	 247
+i_	 247
+na	 245
+ma	 245
+ia	 241
+ar	 234
+sch	 228
+b	 228
+da_	 217
+es	 214
+ai	 213
+st	 212
+'	 209
+e_	 208
+as_	 208
+er_	 200
+t_	 199
+re	 199
+_l	 195
+al	 193
+_n	 193
+el	 192
+tg	 192
+te	 188
+z	 187
+ha	 180
+_f	 180
+sa	 174
+_da_	 172
+ve	 169
+ei	 168
+_v	 165
+at	 165
+ss	 161
+is	 161
+_ch	 161
+on	 160
+la_	 158
+cu	 158
+ad	 158
+he	 156
+in_	 147
+va	 147
+_in	 146
+gl	 145
+ns	 141
+ur	 140
+�	 140
+u_	 138
+ts	 138
+pe	 136
+li	 134
+gi	 133
+et	 133
+de	 132
+ig	 132
+or	 130
+ti	 129
+il	 127
+d_	 126
+che	 123
+ut	 122
+us	 122
+cha	 121
+di	 120
+ia_	 120
+_b	 118
+_la	 117
+na_	 116
+ain	 115
+per	 115
+to	 115
+_cu	 113
+_sc	 112
+se	 111
+ls	 109
+-	 108
+iu	 108
+ca	 107
+si	 104
+ir	 102
+rt	 102
+ie	 102
+_g	 102
+un_	 102
+nd	 101
+av	 101
+ni	 100
+q	 99
+au	 97
+ls_	 97
+ll	 96
+qu	 96
+_pe	 96
+le	 96
+rs	 95
+pa	 95
+ri	 95
+_ma	 94
+_per	 93
+am	 93
+eg	 92
+_r	 92
+gn	 92
+me	 92
+pi	 91
+an_	 91
+en_	 91
+ga	 91
+ent	 90
+hi	 90
+_e_	 90
+it	 89
+ta_	 88
+ter	 87
+ns_	 86
+iv	 86
+igl	 86
+a.	 85
+em	 85
+I	 85
+chi	 84
+_en	 84
+int	 84
+ue	 83
+su	 82
+tt	 82
+a,	 82
+a,_	 82
+o_	 82
+_ve	 82
+a._	 82
+_q	 81
+_qu	 80
+ge	 80
+"	 80
+_la_	 80
+ar_	 80
+vi	 79
+gl_	 79
+tu	 78
+ng	 78
+ro	 76
+mi	 76
+sta	 75
+ed	 75
+lla	 74
+ei_	 73
+_o	 73
+ic	 73
+el_	 73
+_il	 73
+_in_	 72
+g_	 72
+pr	 71
+nu	 70
+ina	 70
+_h	 69
+scha	 68
+mai	 68
+pl	 68
+il_	 68
+os	 68
+ha_	 68
+be	 67
+uo	 67
+cun	 67
+ra_	 67
+_pa	 65
+ts_	 64
+s.	 64
+co	 64
+_u	 64
+fi	 64
+_I	 63
+sa_	 63
+s._	 63
+_re	 62
+�n	 62
+_nu	 62
+?	 62
+S	 62
+_se	 61
+no	 61
+nt_	 61
+E	 61
+tr	 61
+mp	 60
+igl_	 60
+_su	 60
+_st	 60
+ess	 60
+im	 60
+zi	 59
+?_	 59
+nc	 59
+_E	 58
+_-	 58
+_cun	 58
+_te	 58
+_sa	 58
+ant	 58
+main	 58
+eu	 57
+ssa	 57
+-_	 57
+iun	 57
+_-_	 57
+aint	 56
+op	 56
+al_	 56
+dal	 56
+j	 56
+ama	 55
+_tg	 55
+ua	 55
+M	 55
+per_	 55
+tsc	 54
+nta	 54
+tsch	 54
+um	 54
+fa	 54
+za	 54
+_di	 54
+pia	 54
+_per_	 54
+T	 54
+_M	 53
+ne	 53
+era	 53
+A	 53
+que	 53
+_dal	 53
+cr	 53
+_de	 53
+lla_	 53
+_che	 52
+h'	 52
+_me	 51
+ot	 51
+_pr	 51
+_pl	 51
+_sch	 51
+ch'	 50
+_S	 50
+opi	 50
+ou	 50
+tta	 50
+mo	 50
+ada	 50
+ba	 50
+_an	 49
+top	 49
+id	 49
+�	 49
+he_	 49
+C	 49
+va_	 49
+uto	 49
+ins	 48
+topi	 48
+L	 48
+ir_	 48
+ist	 48
+c_	 48
+_il_	 48
+P	 48
+ss_	 48
+ag	 47
+_no	 47
+res	 47
+las	 47
+_vi	 46
+s,	 46
+schi	 46
+_en_	 46
+tg_	 46
+s,_	 46
+_si	 46
+_que	 45
+_T	 45
+az	 45
+'i	 45
+cun_	 44
+_fa	 44
+_mi	 44
+utop	 44
+utopi	 44
+sche	 44
+_C	 44
+ur_	 44
+tge	 44
+po	 44
+es_	 44
+x	 44
+nz	 44
+_L	 44
+_cun_	 44
+man	 44
+_ch'	 43
+_fi	 43
+pu	 43
+ell	 43
+opia	 43
+_igl	 43
+_ig	 43
+sp	 43
+topia	 43
+ava	 42
+egn	 42
+che_	 42
+on_	 42
+ci	 42
+_P	 42
+ev	 42
+ond	 41
+_"	 41
+�	 41
+us_	 41
+_ha	 41
+D	 40
+_co	 40
+etg	 40
+'e	 40
+las_	 40
+est	 40
+ura	 40
+uel	 39
+ed_	 39
+vo	 39
+gia	 39
+bu	 39
+mu	 39
+nn	 39
+gli	 39
+_A	 39
+lu	 39
+cha_	 39
+ul	 38
+mal	 38
+_�n	 38
+ina_	 38
+_bu	 38
+_ca	 38
+_�	 38
+uc	 38
+nts	 38
+tra	 38
+_tu	 37