word alignment 0.00434032
target word 0.0040656500000000005
source word 0.0040223690000000005
word alignments 0.0039668920000000005
word pair 0.003730698
ing word 0.003718359
lexical word 0.0037169800000000004
word level 0.0036928010000000003
single word 0.003635351
word lexicon 0.0036292890000000004
phrase translation 0.00358059
specific word 0.0035607390000000003
word sense 0.003555369
word triplets 0.0035530260000000004
ambiguous word 0.0035426000000000003
empty word 0.0035227110000000004
word graph 0.003505385
successor word 0.003504057
word penalties 0.003504057
spanish word 0.003504057
word senses 0.003504057
language model 0.003216908
machine translation 0.002841048
distortion model 0.002831032
lexical translation 0.00279749
translation probabilities 0.002795826
ibm model 0.002788202
model parameters 0.002761126
model performance 0.002735044
second model 0.002730769
translation process 0.002725994
triplet model 0.002724813
translation tasks 0.002716749
reference translation 0.002676351
model size 0.00267059
translation approaches 0.0026656279999999997
unique translation 0.002665204
overall translation 0.002664162
model con 0.0026574470000000003
lexicon model 0.0026512190000000002
statistical model 0.002634463
translation hypotheses 0.002608982
translation quality 0.002603451
example translation 0.002603304
likely translation 0.002603027
trigram model 0.002601467
translation domain 0.002595457
unconstrained model 0.0025870290000000002
trigger model 0.002586333
translation lexi 0.002584308
proved translation 0.002584308
palign model 0.002534294
model palign 0.002534294
model variations 0.002528603
icon model 0.00252555
translation 0.00234206
model 0.00228348
target words 0.00223419
phrase models 0.001978929
english words 0.001967152
different phrase 0.001940365
training data 0.0019043110000000001
words taxpayer 0.001868504
bilingual phrase 0.0018480810000000001
same phrase 0.001744478
phrase pair 0.001707678
empty words 0.001691251
neighboring words 0.00168629
language models 0.0016738270000000001
phrase level 0.001669781
target sentence 0.001662881
phrase pairs 0.001655049
bilingual training 0.0016550010000000001
current phrase 0.0016169910000000001
training corpus 0.001573176
target source 0.0015649190000000001
phrase selection 0.001527423
previous training 0.00152007
phrase segmentation 0.001512687
gual phrase 0.001483398
phrase segmentations 0.001481668
phrase boundaries 0.001481668
bilingual sentence 0.001468332
different models 0.0014422340000000001
words 0.00143009
alignment constraint 0.001376172
rate training 0.001373319
training corpora 0.001366489
language pairs 0.001349947
training constraints 0.0013428160000000001
alignment path 0.0013373209999999998
training steps 0.001334111
training cor 0.0013287750000000001
sentence pair 0.001327929
final results 0.0013212879999999999
bleu score 0.001321139
iterative training 0.001311796
whole training 0.0013113040000000001
training procedure 0.0013079
evaluation data 0.001292584
