translation model 0.00504892
source word 0.00454592
same word 0.0038323539999999996
word reordering 0.003678746
word order 0.003641531
new translation 0.003552623
word posterior 0.003497742
translation hypotheses 0.0034348219999999997
machine translation 0.0034212929999999997
translation performance 0.003407925
translation probabilities 0.0033923879999999997
word surface 0.0033475849999999997
word reor 0.003343447
translation selection 0.003329982
final translation 0.003288283
original translation 0.003261324
undesirable translation 0.003174308
chine translation 0.003173689
entire translation 0.003173057
translation candi 0.00317031
language model 0.003114847
translation 0.00289766
reordering model 0.002758916
model adaptation 0.00255916
target data 0.00255394
tion model 0.002551866
words data 0.00253952
lation model 0.002497633
source language 0.002438417
mixture model 0.002430869
model self 0.002425342
source sentence 0.002419126
language phrase 0.002339097
test data 0.0021729
training data 0.0021708400000000003
model 0.00215126
target language 0.002091647
different phrase 0.002077528
language words 0.002077227
target sentence 0.002072356
phrase table 0.002056432
source sentences 0.00204223
new phrase 0.002030473
target training 0.00187302
knowledge source 0.0018173050000000002
language models 0.001811392
source input 0.001793055
phrase pairs 0.001777262
open source 0.001774282
source package 0.0017465310000000002
original phrase 0.001739174
bleu score 0.0017278900000000002
phrase segmentation 0.0016518919999999999
new language 0.00161855
monolingual target 0.001511939
source 0.00147483
test set 0.0014263280000000001
words string 0.001389693
phrase 0.00137551
evaluation test 0.001346337
multiple language 0.001343796
smt system 0.00129546
baseline system 0.001286827
english parallel 0.001192352
other feature 0.0011647089999999999
target 0.00112806
system structure 0.001121925
words 0.00111364
posterior probability 0.001110452
original training 0.001108624
combined system 0.001089137
test sets 0.001063934
search strategy 0.001048533
additional feature 0.001036694
learning method 0.0010199319999999999
new list 0.001018936
feature function 0.001017961
development set 0.001009878
probability space 9.89288E-4
iteration number 9.82646E-4
decoding algorithm 9.79327E-4
different kinds 9.75807E-4
similar improvement 9.74295E-4
language 9.63587E-4
first decoding 9.613029999999999E-4
statistical machine 9.53324E-4
sentence 9.44296E-4
performance improvements 9.32427E-4
possible reason 9.31931E-4
posterior probabilities 9.213800000000001E-4
hypotheses list 9.01135E-4
tion probabilities 8.95334E-4
adaptation problem 8.94752E-4
evaluation metric 8.91197E-4
reordering self 8.81738E-4
nist transla 8.81218E-4
score 8.79965E-4
further improvements 8.70907E-4
feature functions 8.65957E-4
results experiments 8.58159E-4
