translation feature 0.00395939
translation features 0.00392407
translation rules 0.00354568
translation forest 0.0033079090000000004
translation probability 0.003281805
reference translation 0.003039488
translation performance 0.003031101
different translation 0.003015686
phrase translation 0.002922259
machine translation 0.00288406
translation scores 0.002861705
translation steps 0.0028350240000000002
high translation 0.0028162580000000003
translation task 0.002810775
reverse translation 0.00279626
corresponding translation 0.002762021
erence translation 0.002756616
cal translation 0.002754013
ence translation 0.002743446
wrong translation 0.002740101
ping translation 0.002740101
translation 0.00247625
word features 0.0024708200000000003
training algorithm 0.00246398
syntactic features 0.002248681
derivation tree 0.00222229
training data 0.002125096
such features 0.00207158
many features 0.002045472
feature weights 0.001901695
context feature 0.001877266
context features 0.0018419460000000001
lexical features 0.001833655
learning algorithm 0.0018284210000000002
traditional feature 0.001825147
source word 0.001813308
language model 0.001794155
traditional features 0.0017898270000000001
new rule 0.001748425
word alignment 0.001744828
such derivation 0.00172682
different training 0.0017256760000000002
overlapping features 0.001711747
model score 0.001674693
new tree 0.001672605
reference derivation 0.001666298
new derivation 0.0016564350000000002
generation algorithm 0.00164138
training example 0.001634217
large data 0.00163077
new rules 0.001622805
rule table 0.001590217
discriminative training 0.001586027
discriminative model 0.001580807
rule scores 0.001580505
training procedure 0.001578109
cky algorithm 0.0015727570000000001
rate training 0.001563142
training sets 0.001553991
reordering rule 0.00154952
algorithm scales 0.0015462940000000001
rule extraction 0.001544425
entire training 0.001537277
source sentence 0.001529835
edge rule 0.001508494
source words 0.0015058329999999998
scfg rule 0.001499046
minimum rules 0.001483467
feature 0.00148314
different data 0.001478292
training times 0.0014706430000000002
tion tree 0.001466797
glue rule 0.001462207
training paradigms 0.0014557530000000002
input tree 0.001452863
sgd training 0.0014513960000000002
training configurations 0.0014513960000000002
training guarantees 0.0014499930000000001
training itera 0.0014499930000000001
features 0.00144782
use tree 0.001435511
word context 0.001417126
context word 0.001417126
such source 0.001414068
traditional rules 0.001411437
rare rules 0.0014030059999999998
small data 0.001398845
allel tree 0.001391672
word sense 0.001373429
scfg rules 0.0013734259999999999
origin derivation 0.001370797
ence derivation 0.001370256
word level 0.001366452
other nodes 0.0013566609999999999
corresponding rules 0.0013552009999999999
reusable rules 0.001351737
word counts 0.001349954
ing data 0.001340308
full data 0.001339593
such derivations 0.001339083
