language_detector 0.1.0

data/lib/textcat_ngrams/swahili.lm

@@ -0,0 +1,400 @@
+_	 16483
+a	 9342
+i	 5293
+a_	 4071
+u	 2730
+k	 2609
+n	 2351
+w	 2076
+m	 1928
+e	 1866
+h	 1800
+o	 1775
+wa	 1743
+l	 1486
+s	 1419
+i_	 1401
+t	 1399
+_k	 1306
+y	 1100
+_w	 969
+li	 945
+wa_	 911
+z	 891
+_wa	 890
+ka	 834
+ku	 799
+r	 770
+b	 733
+an	 727
+ma	 723
+o_	 711
+_m	 707
+na	 698
+ya	 675
+ha	 672
+g	 602
+al	 580
+d	 570
+at	 560
+am	 554
+_n	 549
+_ku	 549
+ik	 543
+_h	 515
+ya_	 514
+A	 494
+is	 487
+_y	 485
+hi	 474
+na_	 471
+_ya	 471
+ta	 468
+sh	 456
+ali	 449
+j	 426
+u_	 423
+ki	 418
+e_	 402
+p	 401
+ti	 401
+_wa_	 399
+f	 392
+_ya_	 390
+ba	 390
+ri	 385
+ng	 385
+il	 380
+c	 358
+hu	 356
+_na	 356
+ni	 355
+za	 354
+zi	 351
+ia	 344
+_na_	 343
+_a	 336
+in	 327
+_ma	 326
+ch	 322
+mb	 317
+ika	 314
+.	 311
+._	 309
+_ka	 307
+as	 306
+ak	 306
+ati	 301
+,	 300
+ka_	 296
+,_	 294
+_u	 292
+kw	 286
+ili	 278
+K	 274
+en	 271
+si	 266
+_kw	 262
+la	 261
+ni_	 261
+ma_	 261
+_s	 258
+kwa	 258
+ar	 256
+ut	 245
+za_	 245
+nd	 242
+mba	 241
+_kwa	 239
+_z	 234
+li_	 233
+un	 233
+ny	 230
+it	 229
+se	 229
+yo	 227
+ia_	 222
+M	 221
+sa	 221
+kat	 217
+_K	 214
+_i	 213
+ika_	 213
+ana	 212
+ish	 212
+kati	 206
+_ha	 204
+on	 201
+ai	 200
+I	 198
+aa	 196
+um	 195
+im	 190
+v	 188
+mu	 187
+amb	 187
+sha	 185
+em	 183
+fa	 181
+zi_	 180
+di	 179
+mi	 178
+_M	 178
+us	 176
+_ki	 176
+ha_	 175
+iw	 172
+ama	 172
+_kat	 168
+_kati	 168
+_hi	 166
+_l	 166
+ra	 166
+kwa_	 165
+la_	 164
+W	 164
+ja	 163
+U	 163
+N	 163
+amba	 161
+ao	 161
+_za	 160
+ji	 160
+B	 157
+iwa	 155
+tik	 155
+wal	 155
+le	 155
+tika	 154
+ge	 153
+lis	 153
+tu	 152
+atika	 152
+to	 152
+atik	 152
+uw	 152
+_kwa_	 151
+A_	 151
+ke	 150
+S	 147
+tika_	 145
+aj	 145
+we	 144
+cha	 144
+bi	 141
+az	 140
+er	 139
+ek	 138
+katik	 138
+ez	 138
+uwa	 137
+kut	 135
+_al	 134
+_B	 134
+ad	 134
+mu_	 133
+_ali	 133
+rik	 132
+_W	 131
+ba_	 131
+kuw	 131
+me	 130
+ali_	 128
+kuwa	 128
+ema	 127
+wan	 127
+bu	 126
+sem	 126
+_A	 125
+ir	 125
+ata	 125
+iz	 124
+_hu	 124
+ay	 124
+ul	 124
+af	 123
+iki	 122
+ema_	 121
+da	 120
+ti_	 120
+sema	 119
+aka	 118
+sema_	 118
+te	 118
+uz	 117
+yo_	 117
+_v	 117
+io	 116
+iy	 115
+uta	 115
+ani	 115
+_wal	 115
+he	 115
+if	 114
+_la	 114
+ab	 114
+go	 112
+_za_	 111
+ama_	 111
+sa_	 111
+pa	 110
+_t	 110
+zo	 110
+nge	 110
+wam	 109
+wali	 108
+ua	 107
+ur	 106
+_c	 106
+ise	 105
+_ch	 105
+isem	 105
+ho	 105
+ye	 104
+iyo	 104
+E	 104
+el	 104
+mo	 103
+ung	 103
+eri	 103
+_wali	 103
+_b	 102
+mba_	 102
+ari	 101
+ita	 101
+isema	 100
+ot	 99
+_la_	 99
+uk	 99
+ao_	 99
+di_	 99
+sha_	 99
+ini	 99
+kuwa_	 98
+uwa_	 98
+ana_	 98
+lise	 98
+lisem	 98
+uli	 97
+shi	 97
+ga	 96
+iwa_	 96
+fu	 96
+T	 96
+R	 95
+_il	 95
+wak	 94
+aw	 94
+isha	 94
+ri_	 93
+_am	 93
+ara	 92
+_cha	 92
+aji	 92
+_ili	 91
+ifa	 91
+O	 90
+_p	 90
+uh	 90
+iri	 90
+chi	 90
+asi	 89
+po	 89
+a.	 89
+ong	 89
+azi	 88
+_j	 88
+_kut	 88
+eny	 88
+nc	 88
+a._	 88
+ko	 87
+uu	 87
+id	 87
+w_	 87
+no	 87
+P	 86
+ah	 86
+ina	 86
+rika	 86
+_Bw	 85
+H	 85
+gu	 85
+uo	 85
+Bw_	 85
+_Bw_	 85
+_se	 85
+Bw	 85
+ib	 84
+_S	 84
+kam	 84
+hi_	 84
+nya	 84
+si_	 83
+a,	 82
+no_	 81
+pi	 81
+ok	 81
+i.	 81
+ip	 81
+kwam	 81
+i._	 81
+amba_	 80
+dh	 80
+end	 80
+ani_	 80
+a,_	 79
+wamb	 79
+kwamb	 79
+_sh	 79
+eza	 79
+nz	 79
+wi	 79
+_kwam	 79
+wamba	 79
+alis	 78
+_kuw	 78
+ngo	 78
+ap	 77
+_N	 77
+any	 77
+ili_	 77
+C	 77
+WA	 76
+vy	 76
+wana	 76
+_hiy	 75
+Wa	 75
+hiyo	 75
+nch	 75
+_hiyo	 75
+de	 75
+_kuwa	 75
+ing	 75
+hiy	 75
+vi	 75
+isha_	 74
+es	 74
+atu	 74
+_Wa	 74
+nchi	 74
+aki	 74
+lim	 73
+da_	 73
+ini_	 73
+ash	 73
+ala	 73
+i,	 73
+ano	 73
+i,_	 72
+_kam	 71
+_wan	 71
+ano_	 71
+mw	 71
+nde	 71
+ji_	 71
+ion	 70
+_amb	 70
+ndi	 70
+_Ka	 70
+eza_	 70

data/lib/textcat_ngrams/swedish.lm

@@ -0,0 +1,400 @@
+_	 33494
+e	 8992
+n	 7900
+t	 7859
+a	 7781
+r	 7251
+s	 6435
+i	 5649
+l	 4541
+d	 4079
+o	 3724
+m	 3203
+k	 3058
+g	 2478
+en	 2403
+n_	 2389
+t_	 2073
+de	 1939
+r_	 1910
+v	 1890
+h	 1789
+u	 1782
+_s	 1768
+�	 1724
+er	 1709
+f	 1597
+en_	 1537
+a_	 1526
+an	 1357
+p	 1320
+et	 1317
+�	 1278
+�	 1261
+st	 1236
+ar	 1226
+c	 1191
+_d	 1158
+e_	 1116
+in	 1045
+_f	 1027
+te	 1000
+b	 997
+_a	 978
+s_	 974
+ra	 958
+.	 956
+tt	 935
+_i	 898
+_m	 890
+._	 886
+ll	 870
+ta	 844
+_o	 842
+_e	 839
+nd	 820
+ti	 804
+sk	 798
+re	 779
+at	 769
+_de	 754
+om	 743
+m_	 739
+�r	 720
+,	 697
+,_	 695
+ng	 686
+li	 673
+ka	 666
+oc	 662
+_h	 654
+on	 652
+et_	 647
+ch	 645
+ns	 643
+is	 642
+er_	 630
+�r	 625
+_v	 614
+_t	 614
+ni	 611
+i_	 609
+_oc	 592
+tt_	 587
+na	 586
+y	 586
+la	 579
+_b	 579
+h_	 577
+kt	 575
+ch_	 568
+ig	 564
+f�	 563
+och	 555
+or	 555
+_och	 554
+och_	 554
+_och_	 553
+me	 548
+den	 548
+om_	 535
+_i_	 531
+d_	 530
+j	 529
+ik	 520
+de_	 520
+f�r	 518
+ge	 498
+ad	 497
+_k	 491
+_f�	 487
+ri	 484
+el	 482
+il	 481
+so	 480
+al	 474
+g_	 469
+le	 464
+an_	 461
+_f�r	 447
+si	 437
+ar_	 437
+att	 435
+_p	 434
+es	 420
+ing	 413
+se	 407
+to	 404
+_u	 403
+_en	 403
+and	 398
+den_	 395
+nde	 393
+nn	 393
+_l	 391
+�_	 391
+D	 385
+�n	 383
+nt	 382
+l_	 381
+tr	 378
+_D	 372
+va	 370
+am	 369
+sa	 367
+_so	 365
+ga	 364
+_en_	 361
+�r_	 358
+ck	 357
+av	 354
+v_	 351
+ed	 347
+ma	 346
+da	 346
+som	 346
+rs	 344
+som_	 344
+ve	 342
+ter	 341
+att_	 341
+ha	 338
+ne	 337
+ut	 335
+as	 332
+ska	 329
+_at	 327
+_att	 326
+_som	 324
+_att_	 324
+_som_	 323
+vi	 322
+ikt	 317
+_av	 316
+det	 316
+_den	 315
+he	 315
+ss	 314
+un	 307
+ke	 304
+_g	 303
+us	 302
+di	 302
+_st	 300
+rn	 297
+_me	 296
+_�	 295
+ade	 294
+"	 290
+_ha	 290
+av_	 289
+ill	 288
+_n	 286
+_in	 279
+io	 275
+_r	 275
+der	 275
+it	 274
+_av_	 274
+sta	 274
+gen	 272
+isk	 270
+_ti	 269
+id	 265
+na_	 265
+ns_	 264
+ko	 262
+_den_	 261
+ag	 258
+det_	 257
+lig	 257
+era	 256
+ll_	 255
+_det	 252
+_�r	 251
+be	 249
+_�r_	 248
+ra_	 247
+ion	 244
+-	 241
+pr	 240
+oni	 233
+til	 231
+ten	 228
+_si	 225
+k_	 222
+p�	 222
+fr	 221
+ro	 219
+till	 219
+iv	 216
+ls	 216
+ande	 215
+�r_	 214
+_det_	 213
+�l	 212
+_p�	 211
+ts	 210
+ens	 209
+med	 209
+mm	 208
+rt	 208
+_till	 208
+_til	 208
+_va	 207
+_fr	 205
+_sk	 205
+var	 205
+nin	 204
+ning	 203
+ol	 201
+ka_	 200
+lle	 198
+ett	 198
+rd	 197
+em	 196
+p�_	 195
+x	 195
+rk	 194
+_ut	 194
+ste	 194
+ds	 193
+_vi	 192
+�r	 192
+S	 192
+nde_	 191
+are	 191
+ver	 190
+_p�_	 190
+nis	 189
+kr	 189
+_med	 188
+all	 188
+�n	 187
+nge	 185
+mo	 184
+os	 183
+ld	 182
+ade_	 181
+_S	 181
+ed_	 180
+r�	 176
+De	 175
+_-	 175
+kan	 174
+ta_	 173
+ng_	 172
+v�	 171
+f�r_	 170
+ill_	 170
+han	 170
+_De	 170
+pp	 169
+lt	 169
+sam	 168
+nte	 167
+ans	 167
+ton	 166
+ur	 165
+mi	 165
+ess	 165
+kl	 164
+ig_	 164
+ks	 164
+as_	 163
+und	 163
+men	 162
+med_	 161
+_med_	 161
+ak	 161
+Di	 160
+ot	 159
+rna	 159
+ul	 159
+_var	 159
+te_	 158
+gen_	 158
+het	 157
+kto	 157
+str	 156
+_Di	 155
+tad	 155
+lan	 154
+ga_	 154
+iska	 154
+fa	 154
+fi	 154
+s�	 154
+Dikt	 153
+Dik	 153
+pe	 153
+ska_	 152
+ja	 152
+H	 151
+res	 151
+ku	 151
+iu	 150
+ande_	 150
+till_	 150
+t.	 150
+ern	 150
+rm	 149
+_Dikt	 149
+_Dik	 149
+ie	 149
+bl	 148
+-_	 147
+od	 147
+_H	 147
+n.	 147
+ist	 147
+_di	 146
+ius	 146
+_"	 145
+la_	 145
+sl	 145
+man	 145
+ren	 145
+_f�r_	 145
+toni	 144
+kton	 144
+n._	 144
+ktoni	 144
+ikton	 144
+I	 144
+ikto	 144
+nius	 143
+ten_	 143
+onius	 143
+oniu	 143
+toniu	 143
+ing_	 143
+Dikto	 143
+niu	 143
+_ko	 143
+ic	 142
+_sa	 142
+_han	 142
+ett_	 142
+sm	 141
+ba	 141
+M	 141
+gr	 140
+l�	 140
+ex	 138
+t._	 138
+sp	 137
+lla	 137
+_et	 137
+_M	 137
+dr	 137
+r�	 136
+rad	 136
+ek	 136
+_be	 135
+tar	 135
+_-_	 135
+_om	 134
+rl	 134
+E	 134
+m�	 133