source word 0.00475774
target word 0.00472224
translation model 0.00456071
language word 0.00453255
phrase translation 0.00401738
large word 0.003976446
word order 0.003942525
word lexicon 0.003909974
asr word 0.003839702
recognition word 0.003825921
single word 0.0038021749999999997
word errors 0.0037953049999999997
word error 0.003790062
input word 0.003787722
translation system 0.00375267
null word 0.0037467029999999997
word graph 0.003725049
word graphs 0.003712854
translation models 0.003710324
word penalty 0.0037079369999999997
separate word 0.003698223
word triphone 0.003694728
word selec 0.0036763869999999997
translation probability 0.0034706750000000003
speech translation 0.0033919280000000002
ibm translation 0.003302088
machine translation 0.003276055
translation results 0.003241116
based translation 0.003099995
the translation 0.003018279
translation cor 0.003014432
automatic translation 0.002994191
oral translation 0.002974463
plex translation 0.002974463
language model 0.00294915
translation 0.00272192
target phrase 0.00259551
model probability 0.002587545
source sentence 0.0025345999999999997
same model 0.0025003250000000003
source language 0.00244591
first model 0.00243516
ibm model 0.002418958
target language 0.00241041
lexicon model 0.002326574
language sentence 0.0023094099999999996
source words 0.002309039
reordering model 0.002308752
general model 0.0022187120000000003
fertility model 0.0021746870000000002
rescoring model 0.0021548220000000002
model scores 0.002150721
different source 0.002139925
acoustic model 0.002131997
individual model 0.0021200990000000003
guage model 0.002115006
language models 0.002098764
alignment models 0.002092424
same source 0.001997085
same target 0.001961585
source text 0.001859579
model 0.00183879
possible target 0.0018219820000000002
different models 0.001792779
language speech 0.001780368
source phrases 0.00175102
baseline system 0.001717948
target phrases 0.0017155200000000001
target part 0.001694132
single target 0.001680035
input target 0.001665582
source toolkit 0.001640698
multiple source 0.001640678
language text 0.001634389
next phrase 0.001629494
phrase pairs 0.001609424
corresponding source 0.001607632
open source 0.001606602
source texts 0.001599319
particular source 0.0015921
phrase penalty 0.001581207
corresponding target 0.001572132
ibm models 0.001568572
test corpus 0.001567115
other words 0.001561611
language sentences 0.0015326709999999998
training corpus 0.001517592
simple language 0.001508979
additional models 0.001497839
text words 0.001497518
language modeling 0.0014779149999999998
lexicon models 0.001476188
cat system 0.001464185
reordering models 0.001458366
asr system 0.001448262
natural language 0.0014469349999999999
recognition system 0.001434481
cat models 0.001421839
previous system 0.001414558
trigram language 0.0014096319999999999
