translation model 0.0039128999999999995
phrase translation 0.0033282109999999998
training data 0.00316657
translation set 0.003107572
translation system 0.0030456729999999996
translation rules 0.002937783
reference translation 0.00289223
translation probabilities 0.0028530919999999998
translation task 0.0028278189999999996
machine translation 0.002821643
lexical translation 0.002772105
translation grammar 0.0027412599999999997
complex translation 0.002693989
translation gram 0.002678205
chine translation 0.002676473
test data 0.002491497
translation 0.00240445
large training 0.002204246
feature set 0.0021992920000000003
language model 0.00217562
model scores 0.002138374
feature weight 0.002101211
other features 0.0020973859999999997
side data 0.002088176
parallel training 0.002035594
dev data 0.002024165
feature vector 0.00201635
distortion features 0.002013521
training sets 0.001967276
data distribution 0.001966854
sparse features 0.001961006
feature sets 0.001946646
testing data 0.001944344
entire data 0.0019359260000000001
many training 0.0019352
training process 0.001933762
data transfer 0.001919176
learning algorithm 0.0019155930000000002
discriminative training 0.001912342
feature selection 0.001888866
dense features 0.0018786319999999999
feature functions 0.001871405
corresponding feature 0.001869189
communication training 0.001853647
feature vectors 0.00183008
scale training 0.0018029790000000001
joint feature 0.00179919
training instances 0.001787665
feature name 0.001785294
feature names 0.001778505
feature spaces 0.001766073
learning method 0.001743806
input sentence 0.00163872
reference translations 0.00162319
bleu test 0.001616662
online learning 0.001600522
source sentence 0.001589835
single learning 0.001589479
features 0.00154872
training 0.0015168
model 0.00150845
feature 0.00149617
structured learning 0.001493056
different source 0.0014905600000000001
hypothesis sentence 0.0014803479999999998
machine learning 0.001463993
learning mira 0.001437026
learning algorithms 0.001421853
learning rate 0.0014160230000000002
ence translations 0.0014086020000000001
large set 0.001390568
batch learning 0.00138316
learning problems 0.0013829710000000002
bleu score 0.001362323
large language 0.001354616
small set 0.001350614
hierarchical phrase 0.00135016
large input 0.001325566
learning curve 0.0013204900000000001
language models 0.0013058549999999999
tuning bleu 0.001300659
test sets 0.001292203
ple sentence 0.001285177
required sentence 0.001269825
inference algorithm 0.001249091
statistics phrase 0.001241553
tuning set 0.001228846
dev test 0.001216122
large tuning 0.00121317
different resources 0.001205508
standard input 0.001203098
different result 0.001184546
bleu ter 0.001183426
different lengths 0.00117604
small tuning 0.001173216
algorithm design 0.001171692
tuning scores 0.001155648
such information 0.001154378
set size 0.001147489
architecture algorithm 0.001141619
