target word 0.00452388
source word 0.00449441
translation translation 0.00415476
word alignment 0.004119388
word class 0.003844454
output word 0.003809585
translation model 0.0037566
word counts 0.0037460419999999998
word classes 0.00373603
word identity 0.003711507
respective word 0.003687853
actual word 0.003679551
word identities 0.003674116
word clusters 0.003670235
translation models 0.00312501
language model 0.00269346
ibm translation 0.002588814
translation task 0.0024872600000000003
machine translation 0.002485526
analysis translation 0.002473766
translation generation 0.002422322
translation quality 0.0024055400000000003
german translation 0.002401105
translation probabilities 0.002399618
translation systems 0.002382901
translation tasks 0.0023798770000000003
translation process 0.002365044
translation framework 0.0023541540000000002
frequency translation 0.002335497
translation paradigms 0.0023347980000000003
ferent translation 0.002329574
translation decoders 0.002323243
translation directions 0.0023191440000000004
factored translation 0.0023191360000000003
translation paradigm 0.0023166790000000003
translation option 0.002312577
translation koehn 0.002312577
same model 0.002208047
training data 0.002120182
reordering model 0.002082755
translation 0.00207738
standard model 0.002036721
model type 0.001987654
model estimation 0.001970407
separate model 0.001960562
hrm model 0.001915349
model files 0.001914344
model file 0.001914344
target sentence 0.001914299
source sentence 0.0018848289999999998
phrase segmentation 0.001858759
training corpus 0.001755224
same phrase 0.0017438570000000001
same training 0.001743167
bilingual training 0.0017127169999999999
model 0.00167922
phrase table 0.001644654
corresponding phrase 0.001630281
phrase pair 0.0015675749999999999
data set 0.001540305
training procedure 0.0015355389999999998
feature score 0.001531288
rate training 0.001522565
clustering words 0.001515625
target class 0.001498734
few training 0.001494099
joint phrase 0.001484176
additional language 0.001474276
class models 0.0014572839999999999
gual training 0.001456766
huge training 0.0014522139999999998
phrase expansion 0.001452095
reordering models 0.001451165
training instances 0.001449857
same data 0.001434669
target side 0.001429843
test bleu 0.001411406
source toolkit 0.0014108659999999998
feature weights 0.001410689
standard models 0.001405131
train models 0.001400596
count models 0.001357381
feature combination 0.0013543449999999999
target factors 0.0013460260000000002
channel models 0.001336177
arbitrary models 0.001313164
original models 0.001312916
natural language 0.001309852
language process 0.001301904
open source 0.0012947879999999998
above models 0.001284005
hrm models 0.001283759
words 0.00127938
language processing 0.001263884
portant feature 0.0012543569999999998
additonal feature 0.001239877
ter bleu 0.001233292
bleu ter 0.001233292
phrase 0.00121503
training 0.00121434
