error training 0.00328595
word error 0.00322302
translation model 0.0031139899999999996
different error 0.003011549
different translation 0.002837139
human translation 0.002759083
translation evaluation 0.002726499
translation approach 0.002698759
translation models 0.002697474
english translation 0.002622795
translation probability 0.0026013399999999997
translation task 0.002583852
minimum error 0.002573064
error criterion 0.0025373130000000002
error measure 0.002503559
error count 0.00250096
machine translation 0.0024824829999999997
error counts 0.002482159
translation quality 0.002479976
translation system 0.002472353
error rate 0.00247131
ing error 0.0024704
error criteria 0.002461811
certain error 0.002446611
translation results 0.00244398
classification error 0.002427495
minimal error 0.002400533
unsmoothed error 0.002398812
baseline translation 0.002393792
independent error 0.002387909
error rates 0.002375927
error crite 0.002371627
sification error 0.002362722
reference translation 0.002343203
candidate translation 0.0023430919999999997
direct translation 0.0023356979999999998
hypothesis translation 0.002330691
good translation 0.0022718729999999998
final translation 0.002270974
translation length 0.002246019
different model 0.002119129
different training 0.0021166789999999998
error 0.00209041
language model 0.002067407
training corpus 0.002058909
training data 0.002039554
word errors 0.002032482
training algorithm 0.0019502489999999998
translation 0.001916
test corpus 0.001838136
test data 0.001818781
test sentences 0.00180694
line optimization 0.001783611
various training 0.001773844
new model 0.001765075
new training 0.0017626249999999999
different sentences 0.001753312
different evaluation 0.001731638
different approach 0.001703898
word alignment 0.001702305
model parameter 0.0017015989999999998
new line 0.001655165
training criterion 0.001642443
model parameters 0.001636203
training problem 0.001620528
corresponding training 0.001602814
rate training 0.00157644
different decision 0.001576094
source sentence 0.001574435
alternative training 0.001570409
training criteria 0.0015669409999999999
phrase penalty 0.001560091
feature function 0.001558685
decision errors 0.001554827
training procedure 0.00154803
possible translations 0.001536641
reference sentence 0.0015129129999999998
probabilistic model 0.001504547
input sentence 0.0014999459999999998
training mwer 0.0014958699999999998
tive training 0.0014845079999999998
mmi training 0.001480656
loss function 0.001480191
training crite 0.001476757
model param 0.0014709369999999999
cient training 0.001468023
different features 0.0014515869999999998
optimization algorithm 0.00145024
different results 0.001449119
output sentence 0.0014389469999999999
word order 0.001436858
efficient line 0.0014357719999999999
statistical approach 0.0014125549999999998
line min 0.00141133
target language 0.00139266
bleu score 0.0013780139999999999
probable sentence 0.001363308
line optimiza 0.001362573
ery sentence 0.0013592259999999998
target sentences 0.001355416
