hypothesis word 0.004237305
word order 0.00423194
different word 0.004181596
word state 0.004174711
word position 0.004103074
word sequence 0.00404687
word align 0.004009434
empty word 0.004001752
real word 0.003933486
word sequences 0.003910985
word con 0.0038935999999999997
alternative word 0.00388676
certain word 0.0038798089999999997
word repre 0.0038652979999999997
pothesis word 0.003859779
word penalty 0.0038595749999999996
tual word 0.00385524
ternative word 0.00385524
alignment method 0.003417616
ihmm alignment 0.003381991
hypothesis alignment 0.003334665
alignment methods 0.0033081680000000002
optimal alignment 0.0031106180000000003
alignment process 0.0030833370000000002
incremental alignment 0.003066061
ter alignment 0.003061625
alignment strategy 0.003054097
alignment normalization 0.003051402
alignment assumption 0.0030457540000000003
alignment approaches 0.0030111170000000002
round alignment 0.002967443
monotonic alignment 0.002958811
alignment a˜k 0.0029572720000000004
timal alignment 0.0029571600000000003
tal alignment 0.002953872
alignment 0.00268554
ihmm model 0.0021286309999999998
language model 0.002064102
distortion model 0.0020345909999999997
similarity model 0.002005954
model score 0.001935001
hypothesis translation 0.001894715
other words 0.00186271
hypothesis words 0.001857845
ment model 0.001810394
markov model 0.0017670639999999998
incihmm model 0.001712882
same system 0.001711127
tortion model 0.001702828
translation probabilities 0.0016941579999999999
distortion probability 0.001670691
system order 0.00164979
empty words 0.001622292
lexical translation 0.001607133
different system 0.001599446
machine translation 0.00157455
translation edit 0.0015695029999999999
component translation 0.001564801
training data 0.001556972
source sentence 0.001551798
certain translation 0.001537219
translation quality 0.001528808
dividual translation 0.001512798
native words 0.00147639
way system 0.001467112
transition probability 0.001460462
system combination 0.001445908
model 0.00143218
ihmm method 0.001428527
final system 0.0014250769999999999
viterbi probability 0.001422852
tion probability 0.001416179
next system 0.001356033
emission probability 0.001349406
decoding algorithm 0.0013492629999999999
single system 0.001347107
ihmm models 0.001345545
conventional system 0.001320426
risk system 0.001312263
ihmm distortion 0.001298862
ihmm approach 0.001296687
reversal system 0.0012946939999999999
system combina 0.001278389
same performance 0.001277711
same backbone 0.001254354
distortion models 0.001251505
translation 0.00124559
bleu training 0.001239147
different order 0.001237176
words 0.00120872
such transition 0.0011981049999999999
same way 0.001166179
programming algorithm 0.001159616
target language 0.001153832
first transition 0.001138716
other hypotheses 0.001137538
test set 0.001124468
ter method 0.001108161
same transition 0.001097279
first solution 0.001093316
