word alignment 0.00499397
word translation 0.00475795
translation model 0.0041871
word alignments 0.00378293
alignment models 0.0035822
translation models 0.00334618
english alignment 0.00313462
english word 0.00313381
hmm word 0.003034671
english translation 0.0028986
alignment error 0.002846279
alignment quality 0.002843568
word translations 0.002829435
language model 0.00282895
word type 0.00282238
translation table 0.002821595
translation performance 0.002813609
distinct word 0.002803423
word trans 0.002796791
french word 0.002779625
alignment direction 0.002778726
word classes 0.0027772
automatic word 0.002772124
alignment accuracy 0.002762523
alignment precision 0.00275239
glish word 0.002746823
translation rules 0.002717891
machine translation 0.002705678
translation system 0.002703708
translation experiments 0.00267064
translation quality 0.002607548
translation systems 0.002585853
statistical translation 0.002579049
good translation 0.002576488
translation workshop 0.002562411
translation prob 0.002561252
ibm model 0.00253326
translation accuracy 0.002526503
translation sys 0.00251942
translation pipelines 0.002512003
alignment 0.00249739
hmm model 0.002463821
tion model 0.002355694
translation 0.00226137
language models 0.00198803
model 0.00192573
source words 0.0018238899999999999
baseline alignments 0.001711017
ibm models 0.00169234
standard alignments 0.00168725
same data 0.001685663
parallel data 0.00161469
different language 0.001611047
symmetrized alignments 0.0015773979999999998
other models 0.0015765000000000002
extra alignments 0.001563901
gold alignments 0.0015573179999999998
many models 0.001547859
task data 0.001480767
other language 0.00139491
english sentence 0.0013918
training method 0.001376404
language pairs 0.0013657930000000001
available data 0.001347762
distortion models 0.001338847
development data 0.001324393
newsgroup data 0.001303722
language pair 0.00129081
alignments 0.00128635
training set 0.001256145
rare words 0.001226924
source extension 0.001217392
search algorithm 0.001214873
discriminative training 0.001196145
unrelated words 0.001174992
lion words 0.001171926
uncommon words 0.001171184
likelihood training 0.001166064
ferent language 0.0011636020000000001
standard training 0.001155351
ond language 0.001153505
models 0.00108481
same number 0.0010747
heuristic parameters 0.00105291
ibm bleu 0.001052769
training procedure 0.001042147
efficient training 0.001038902
french sentence 0.001037615
baseline ibm 0.0010321969999999999
features 9.94708E-4
test set 9.9403E-4
objective function 9.87151E-4
english string 9.81134E-4
previous work 9.71561E-4
feature weights 9.55872E-4
search next 9.55074E-4
many parameters 9.51791E-4
large datasets 9.39235E-4
same setting 9.31468E-4
ibm mod 9.25673E-4
