language_detector 0.1.0

data/lib/textcat_ngrams/danish.lm

@@ -0,0 +1,400 @@
+_	 21274
+e	 9291
+r	 5307
+n	 4733
+i	 3976
+t	 3948
+s	 3751
+a	 3296
+l	 3063
+d	 3025
+o	 2868
+g	 2471
+er	 2164
+k	 2002
+m	 1680
+e_	 1655
+en	 1613
+f	 1507
+de	 1484
+r_	 1379
+v	 1245
+u	 1176
+t_	 1081
+n_	 1032
+er_	 992
+b	 942
+.	 870
+ge	 868
+._	 831
+re	 816
+h	 816
+et	 813
+te	 813
+p	 806
+in	 788
+or	 775
+_s	 753
+_a	 749
+en_	 712
+_e	 691
+ti	 689
+an	 687
+,	 681
+,_	 677
+_f	 655
+_d	 645
+el	 642
+ng	 635
+nd	 634
+g_	 634
+se	 615
+le	 615
+st	 607
+s_	 601
+_o	 572
+ne	 560
+li	 537
+et_	 524
+es	 521
+_i	 512
+ri	 511
+sk	 510
+_de	 498
+�	 497
+ar	 475
+ed	 473
+ig	 463
+at	 452
+_m	 446
+is	 443
+fo	 441
+�	 441
+ve	 438
+_k	 434
+�	 432
+der	 429
+ke	 428
+ing	 427
+og	 426
+_b	 412
+me	 408
+il	 407
+for	 405
+ns	 394
+y	 389
+_h	 380
+_t	 374
+on	 371
+d_	 370
+al	 362
+be	 359
+_fo	 351
+af	 336
+de_	 335
+_og	 333
+_p	 332
+og_	 325
+om	 325
+_for	 324
+_og_	 313
+l_	 308
+nge	 302
+i_	 295
+_v	 294
+c	 289
+ter	 283
+ll	 280
+ni	 278
+nde	 278
+rs	 277
+_af	 277
+un	 275
+ra	 271
+ko	 271
+den	 270
+_i_	 268
+id	 265
+til	 265
+j	 265
+vi	 264
+D	 260
+ere	 256
+ma	 255
+si	 253
+f_	 252
+af_	 238
+_af_	 235
+ik	 235
+m_	 234
+�_	 232
+_ti	 227
+_D	 226
+_u	 226
+_er	 225
+nt	 224
+_en	 224
+ls	 221
+es_	 216
+lig	 216
+ger	 216
+re_	 210
+ag	 210
+_me	 207
+at_	 204
+lle	 200
+ge_	 200
+_til	 200
+ige	 199
+_er_	 199
+der_	 199
+em	 199
+ds	 197
+r.	 195
+io	 195
+r._	 195
+ud	 193
+_at	 192
+_at_	 191
+ta	 190
+els	 190
+_l	 190
+ha	 190
+il_	 189
+or_	 189
+ke_	 186
+rt	 185
+gen	 184
+ka	 183
+-	 180
+rk	 180
+ning	 178
+ol	 178
+nin	 178
+la	 177
+ld	 175
+De	 175
+it	 173
+ede	 172
+ed_	 171
+_ko	 171
+lse	 171
+ek	 168
+else	 167
+inge	 167
+p�	 167
+ng_	 167
+_p�	 167
+iv	 166
+�r	 166
+so	 165
+he	 165
+ens	 165
+ske	 165
+ind	 164
+til_	 163
+rn	 163
+ide	 162
+ev	 162
+den_	 162
+to	 162
+sen	 160
+_be	 160
+sa	 160
+bl	 158
+_g	 158
+an_	 157
+det	 156
+om_	 156
+ru	 156
+va	 155
+_til_	 155
+ste	 154
+rd	 153
+_p�_	 152
+k_	 152
+p�_	 152
+di	 152
+kr	 152
+K	 151
+_De	 149
+for_	 148
+te_	 148
+kon	 148
+ver	 147
+mm	 146
+am	 146
+_en_	 145
+_r	 145
+ne_	 144
+ing_	 144
+tr	 143
+le_	 142
+del	 142
+_in	 142
+gt	 140
+_st	 138
+S	 138
+eg	 138
+gs	 138
+tt	 138
+r,	 137
+ser	 137
+r,_	 137
+er.	 137
+ro	 137
+er._	 137
+_for_	 136
+ent	 136
+kt	 136
+eri	 135
+ur	 134
+lin	 134
+B	 133
+A	 133
+sti	 133
+ner	 133
+da	 133
+ris	 132
+ion	 132
+_K	 131
+ern	 131
+ers	 130
+ist	 130
+�r	 130
+ige_	 130
+_si	 130
+tte	 129
+E	 128
+_n	 128
+nn	 127
+_B	 126
+_ha	 126
+_.	 126
+rne	 125
+H	 125
+_ud	 125
+rin	 124
+na	 124
+und	 124
+ft	 124
+_der	 124
+ku	 123
+_A	 122
+ler	 120
+and	 120
+end	 120
+ns_	 120
+rg	 119
+op	 119
+er,_	 119
+er,	 119
+ar_	 118
+P	 118
+_S	 117
+_H	 117
+_._	 116
+ov	 116
+erne	 115
+tio	 115
+med	 115
+tion	 115
+_E	 115
+_P	 115
+det_	 114
+pr	 114
+e.	 113
+ter_	 113
+:	 113
+kk	 113
+e._	 113
+e,_	 113
+e,	 113
+od	 113
+kke	 113
+ten	 113
+ling	 113
+:_	 112
+mi	 112
+eli	 112
+lo	 111
+som	 111
+_den	 111
+rb	 110
+se_	 110
+ell	 110
+sid	 110
+nne	 109
+fi	 108
+lt	 107
+v_	 107
+_de_	 107
+ark	 106
+lige	 106
+ngen	 106
+ie	 105
+_med	 105
+_der_	 105
+ring	 105
+a_	 105
+_vi	 104
+-_	 104
+ys	 103
+gel	 103
+_so	 103
+ia	 103
+ive	 102
+ej	 101
+ati	 101
+ren	 101
+_det	 101
+side	 101
+ske_	 101
+br	 100
+gi	 100
+F	 100
+M	 100
+ul	 99
+isk	 99
+men	 99
+n,_	 99
+age	 99
+fr	 99
+n,	 99
+tu	 98
+ts	 98
+_ma	 98
+nder	 98
+ot	 97
+dt	 97
+R	 97
+med_	 96
+ho	 96
+ans	 95
+_kon	 95
+pe	 95
+ce	 94
+gr	 93
+mme	 92
+ret	 92
+lige_	 92
+mu	 91
+_med_	 91
+hv	 91
+v�	 91
+Det	 91
+ens_	 91
+kl	 91
+_M	 90
+T	 90
+ingen	 90
+rm	 90
+ill	 89
+elle	 89
+ef	 89
+ene	 89
+nds	 89
+ove	 89
+som_	 89
+C	 88
+_den_	 88

data/lib/textcat_ngrams/dutch.lm

@@ -0,0 +1,400 @@
+_	 20104
+e	 9848
+n	 5323
+a	 3733
+t	 3683
+i	 3490
+r	 3195
+d	 2876
+o	 2845
+n_	 2443
+en	 2439
+s	 2195
+e_	 1842
+l	 1837
+g	 1522
+en_	 1500
+de	 1489
+er	 1388
+t_	 1377
+v	 1253
+u	 1217
+k	 1204
+_d	 1136
+h	 1102
+m	 1084
+an	 939
+te	 875
+j	 857
+in	 810
+_v	 793
+r_	 751
+de_	 742
+ee	 737
+p	 732
+et	 718
+ge	 716
+aa	 708
+b	 703
+_e	 686
+st	 669
+z	 668
+ie	 662
+_de	 655
+w	 631
+c	 611
+.	 604
+s_	 582
+_de_	 576
+_h	 572
+el	 570
+ij	 564
+._	 554
+et_	 531
+an_	 522
+he	 505
+_o	 497
+nd	 478
+_i	 475
+ar	 459
+_m	 451
+re	 442
+ve	 441
+'	 428
+or	 424
+ng	 421
+at	 418
+_s	 415
+oo	 403
+_z	 401
+le	 395
+_b	 394
+_a	 391
+_he	 386
+va	 385
+er_	 381
+me	 372
+_w	 368
+f	 361
+on	 351
+_t	 351
+_va	 345
+_g	 342
+di	 342
+nt	 340
+,	 335
+g_	 335
+,_	 334
+van	 327
+ch	 326
+is	 326
+ing	 325
+be	 325
+ni	 320
+it	 317
+een	 316
+_van	 315
+al	 310
+den	 309
+ti	 309
+van_	 307
+oe	 302
+ke	 302
+_van_	 299
+aar	 299
+d_	 295
+we	 293
+da	 292
+tu	 290
+_ee	 290
+ud	 287
+een_	 286
+li	 284
+es	 282
+_st	 281
+ver	 281
+ten	 281
+ri	 275
+nde	 275
+der	 274
+_in	 270
+k_	 268
+vo	 267
+het	 266
+oor	 264
+_het	 262
+het_	 262
+_het_	 259
+_een	 258
+l_	 258
+ze	 257
+_n	 254
+ro	 248
+gen	 243
+_een_	 241
+at_	 240
+op	 238
+n.	 238
+_en	 237
+rs	 237
+_da	 235
+stu	 232
+in_	 230
+_be	 229
+_ge	 228
+_k	 226
+rd	 226
+tud	 220
+_en_	 220
+n._	 217
+te_	 209
+ei	 208
+ent	 206
+_me	 203
+la	 202
+ek	 202
+ed	 201
+ra	 200
+stud	 200
+en.	 200
+ie_	 197
+ste	 196
+_vo	 195
+_in_	 193
+_stu	 191
+zi	 191
+om	 189
+ui	 189
+en._	 186
+ten_	 185
+_stud	 185
+ude	 184
+die	 183
+ns	 183
+_j	 181
+D	 179
+aan	 179
+se	 179
+ma	 178
+_ve	 176
+ne	 174
+_p	 174
+eg	 173
+p_	 172
+ar_	 172
+aar_	 171
+_te	 170
+ng_	 169
+_we	 169
+''	 167
+_D	 165
+ers	 164
+_op	 163
+dat	 161
+dat_	 160
+ig	 160
+ere	 159
+eer	 158
+_zi	 158
+voor	 156
+voo	 156
+nge	 155
+nder	 151
+nte	 151
+or_	 150
+ta	 150
+je	 149
+ing_	 148
+ll	 148
+_ver	 147
+jk	 146
+oor_	 146
+_dat	 145
+ijk	 145
+ren	 145
+is_	 145
+_dat_	 144
+_l	 144
+and	 144
+lij	 143
+ter	 143
+na	 142
+uden	 139
+tude	 138
+_voor	 136
+_voo	 136
+ond	 136
+ken	 135
+cht	 135
+_al	 135
+ht	 135
+wa	 134
+ho	 133
+em	 133
+den_	 133
+pe	 132
+sc	 132
+un	 131
+ur	 131
+_di	 130
+gen_	 130
+zo	 129
+rt	 129
+ev	 128
+mo	 128
+lijk	 127
+_is	 126
+stude	 124
+ha	 123
+to	 122
+el_	 121
+og	 121
+op_	 121
+sch	 120
+ol	 120
+ente	 119
+_u	 118
+pr	 118
+end	 118
+mi	 117
+iet	 116
+_aa	 116
+eli	 115
+dent	 115
+ijn	 115
+jn	 115
+ou	 115
+men	 114
+_'	 114
+tie	 113
+_is_	 113
+nie	 113
+tr	 112
+ak	 112
+id	 112
+udent	 111
+tuden	 111
+uit	 110
+_te_	 109
+aan_	 109
+ld	 109
+S	 108
+_aan	 108
+ede	 108
+ja	 107
+nten	 107
+it_	 107
+je_	 107
+ts	 107
+erd	 106
+est	 106
+E	 105
+_op_	 105
+ad	 104
+al_	 104
+_ze	 104
+_on	 104
+rk	 104
+lle	 103
+ens	 103
+gel	 103
+m_	 103
+len	 103
+_r	 102
+ec	 102
+inge	 102
+met	 102
+_met	 101
+si	 100
+die_	 100
+us	 100
+onde	 99
+_ni	 99
+De	 99
+eu	 99
+dente	 99
+enten	 99
+ic	 99
+_met_	 98
+f_	 98
+met_	 98
+no	 97
+ko	 96
+voor_	 96
+rde	 96
+H	 96
+ngen	 95
+lo	 95
+ot	 95
+as	 94
+zij	 93
+_nie	 92
+vi	 92
+eb	 92
+_De	 92
+_zij	 91
+ep	 91
+wi	 91
+_zo	 91
+kt	 91
+ege	 91
+G	 91
+bi	 90
+j_	 90
+ij_	 90
+ze_	 90
+do	 90
+lan	 89
+ov	 89
+udi	 89
+ord	 89
+onder	 89
+V	 88
+elij	 88
+_wa	 88
+elijk	 88
+ef	 88
+_die	 87
+ag	 86
+erk	 86
+eren	 86
+R	 85
+ik	 85
+_ma	 85
+gr	 85
+am	 85
+_mo	 84
+ul	 84
+nn	 83
+eve	 83
+De_	 83
+maa	 83
+ingen	 83
+wo	 83
+_''	 83
+O	 83
+tudi	 82
+I	 82
+nt_	 82
+tudie	 81
+ven	 81
+udie	 81
+nten_	 81
+_die_	 81
+jaa	 80
+ka	 80
+eke	 80
+ite	 80
+a_	 80
+_je	 80
+ac	 80
+jaar	 80
+_je_	 79
+_H	 79
+_zijn	 79
+zijn	 79
+n,	 78
+nen	 78
+N	 78
+n,_	 78
+ijn_	 77