phrase translation 0.00602517
translation model 0.00473583
phrase pair 0.003961442
phrase pairs 0.0038867479999999998
translation probability 0.003874866
phrase table 0.003797056
translation probabilities 0.0035718879999999996
phrase extraction 0.003553652
machine translation 0.0035387229999999997
translation performance 0.003491299
phrase count 0.003422287
correct phrase 0.00339034
translation knowledge 0.0033883819999999997
translation modeling 0.003368646
vertices phrase 0.003359935
frequent phrase 0.003351808
inverse phrase 0.003340155
conventional phrase 0.003330591
incorrect translation 0.0033196809999999997
translation tasks 0.0033080479999999996
literal translation 0.0033014539999999997
phrase 0.00302637
language model 0.00301174
translation 0.0029988
data table 0.002791626
parallel data 0.002779707
other data 0.002763255
bilingual data 0.002621584
data quality 0.002414391
quality data 0.002414391
development data 0.002398967
noisy data 0.002369191
testing data 0.0023655100000000004
web data 0.002336434
data cleaning 0.002334039
allel data 0.0023222340000000003
gual data 0.002322179
data mining 0.0023190150000000002
data detection 0.002315803
word alignment 0.002300828
model size 0.002205409
reordering model 0.002189813
linear model 0.002147749
programming model 0.002049453
chinese word 0.001905258
sentence pair 0.001896372
sentence pairs 0.001821678
word count 0.0017824569999999999
labeled training 0.001773833
training instances 0.001771825
parallel corpus 0.001763257
model 0.00173703
parallel sentence 0.001720067
different machines 0.0017042329999999999
symmetric word 0.0017015709999999998
different proportions 0.001697358
ing method 0.001654977
sentence length 0.001606255
bilingual corpus 0.001605134
sentence scores 0.0015899590000000002
corpus weighting 0.0015697340000000001
bilingual sentence 0.001561944
baseline system 0.0015123369999999999
pair probabilities 0.0015081600000000001
unsupervised method 0.001478901
method substan 0.001456176
tion probability 0.001431437
good sentence 0.0014242600000000001
training 0.0014146
smt system 0.0013710390000000001
pair frequency 0.001284517
pair vertex 0.001278249
learning methods 0.0012753299999999999
language 0.00127471
pair vertices 0.001268637
long sentence 0.001257803
vertex set 0.001254939
tence pair 0.001238688
pair ver 0.001232164
test nist 0.001218795
same key 0.0012128310000000002
experimental results 0.001205565
table smoothing 0.001193114
smt systems 0.001185358
significance test 0.001178814
pairs usu 0.001169902
random walk 0.001169686
weighting feature 0.001168393
tence pairs 0.001163994
ldc english 0.00116261
method 0.00115946
open test 0.0011409130000000001
test cwmt 0.001135031
tal results 0.00111165
nations parallel 0.001102052
large bilin 0.001100252
random sampling 0.00109497
english part 0.001092266
other hand 0.001080591
previous work 0.001078137
