translation model 0.00401918
source word 0.00307784
phrase translation 0.002999698
translation system 0.002981377
target word 0.00298067
language model 0.00285305
word lattice 0.002809812
machine translation 0.0027237909999999997
possible translation 0.002688936
translation probabilities 0.002667219
morphological feature 0.00265346
translation length 0.002636785
translation space 0.002594903
unsegmented translation 0.002586617
translation quality 0.002573807
word sequence 0.002553444
finnish translation 0.002528233
word alignment 0.002490615
translation outputs 0.002488487
translation pipeline 0.002480736
morphological features 0.002467725
expressive translation 0.002464938
word boundary 0.002417991
word graph 0.002408415
word forms 0.002382121
desegmented word 0.002381471
entire word 0.002338327
word contexts 0.002293916
word boundaries 0.002291517
word bsyarth 0.002286892
unrelated word 0.002286892
spect word 0.002286892
morphological segmentation 0.002281094
translation 0.00222515
model sparsity 0.002158139
unsegmented model 0.002155497
reordering model 0.002154881
model fea 0.002147739
model estimation 0.002135211
model parameters 0.002106972
linear model 0.002103363
source language 0.00209027
factored model 0.0020806080000000003
model dev 0.002042659
morphological analysis 0.002040988
target language 0.0019931
morphological seg 0.001923719
morphological segmenter 0.001898959
source words 0.001861495
morphological complexity 0.00183367
morphological pieces 0.001819671
source phrase 0.001805798
source context 0.001795546
model 0.00179403
target words 0.001764325
language models 0.001661693
target segmentation 0.001636514
feature set 0.001621866
first feature 0.0015944700000000002
standard feature 0.001574723
feature layer 0.001546302
source text 0.001519047
training data 0.001490375
additional feature 0.001480062
segmented language 0.0014450259999999999
target text 0.001421877
unsegmented language 0.0014204869999999998
data set 0.001415594
source phrases 0.001414412
language modeling 0.001408997
arabic training 0.001392699
output lattice 0.001389758
ing target 0.001389602
particular language 0.0013880709999999998
major source 0.0013868910000000001
same stem 0.001379211
target morpheme 0.001363658
language pair 0.001363651
feature vectors 0.001363595
feature ablation 0.0013506640000000001
target languages 0.001347349
mented language 0.001346931
target mor 0.001344184
source tokens 0.001340882
source lan 0.001327883
segmentation baseline 0.001324493
discontiguous source 0.001321618
abstract feature 0.0013206630000000001
segmented target 0.001320086
target phrases 0.001317242
feature groups 0.001315851
target hypothesis 0.00131046
parallel data 0.001305563
segmentation rules 0.0013005920000000002
single target 0.001295796
unsegmented target 0.001295547
arabic desegmentation 0.0012897569999999999
tiguous source 0.0012845600000000001
lattice desegmentation 0.001282127
desegmentation lattice 0.001282127
