word training 0.004513022
segmented word 0.004188174
surface word 0.004149871
word level 0.004149292
word ambiguity 0.0040423030000000006
base word 0.003979788
word penalty 0.003977795
word wlawladh 0.003977372
word changes 0.003975172
resolve word 0.003975172
face word 0.003975172
word lkn 0.003975172
translation model 0.00339748
arabic stem 0.0032814050000000003
arabic corpus 0.00323132
arabic words 0.0031808640000000003
arabic source 0.003110161
arabic segmentation 0.0030518439999999997
arabic target 0.0030233079999999997
language model 0.00299186
segmented arabic 0.002895724
arabic surface 0.0028574209999999997
arabic side 0.002750044
ldc arabic 0.002687775
arabic gigaword 0.0026841169999999998
arabic exper 0.002682732
phrase translation 0.0025640100000000002
arabic 0.0023682
translation models 0.002345893
tion model 0.002341171
distortion model 0.002270149
model orders 0.002224655
translation table 0.002122735
machine translation 0.001985293
language models 0.0019402730000000002
factored translation 0.001925042
sentence test 0.00192208
model 0.00189577
different language 0.001845804
chine translation 0.0018161520000000001
language domain 0.001762202
morphological segmentation 0.001736454
phrase length 0.001620392
training data 0.001613459
test set 0.001605617
translation 0.00150171
sentence length 0.0014898210000000001
morphological decomposition 0.001468673
training set 0.0014676379999999998
trigram language 0.0014462540000000001
morphological analyzer 0.001434233
morphological decompo 0.0014308839999999999
unigram language 0.001426256
morphological fea 0.001405779
english source 0.0013876000000000001
corpus size 0.001387248
imum phrase 0.00137874
news training 0.001361506
such features 0.001322447
baseline system 0.0013107589999999999
surface words 0.001301885
english target 0.001300747
average sentence 0.001290134
tion models 0.0012895839999999999
sentence tuning 0.001285093
factored models 0.0012675149999999999
smt system 0.0012569109999999999
erage sentence 0.001247104
factored words 0.001235996
same size 0.001226861
scheme training 0.0012187230000000001
models experiments 0.001198885
english statistical 0.001192459
ing words 0.001186331
glish corpus 0.001182396
gigaword corpus 0.001179037
guage models 0.0011783829999999999
models experiment 0.0011589719999999999
prior work 0.0011574950000000001
previous work 0.001143767
english surface 0.0011348600000000001
english smt 0.001127138
contextual information 0.0011170870000000001
segmented form 0.0011115320000000001
system moses 0.001111308
moses system 0.001111308
context analysis 0.0011104069999999999
grammatical case 0.001106202
language 0.00109609
travel domain 0.001085574
different sources 0.001073285
surface form 0.001073229
linguistic information 0.0010702490000000001
different recombina 0.0010669639999999999
phrase 0.0010623
source benefits 0.0010566809999999999
full set 0.001049247
segmentation decrease 0.001049178
related work 0.001047998
target side 0.0010369519999999998
