source word 0.00401082
word source 0.00401082
target word 0.00366837
word alignment 0.003304301
translation model 0.00327088
unsupervised word 0.00300452
arabic word 0.0029879060000000002
word embedding 0.002929327
word count 0.002911251
word classes 0.002899555
word tokenization 0.002889903
word clustering 0.002886725
context model 0.002756923
language model 0.002738935
source words 0.00271442
network model 0.0024630299999999997
probability model 0.00246126
language translation 0.002412015
target words 0.00237197
source context 0.002317603
standard model 0.002274284
nnltm model 0.002104476
joint model 0.002089124
model variations 0.002074805
target phrase 0.00205786
lexical translation 0.001986768
target context 0.001975153
machine translation 0.001960665
target language 0.001957165
additional source 0.001874498
source representation 0.0018696379999999999
bleu bleu 0.001834706
model 0.0017989
translation modeling 0.0017749349999999998
vocabulary words 0.001772543
translation joint 0.0017622039999999999
language models 0.001739161
source window 0.0017059689999999999
training data 0.001704624
multiple source 0.001692951
unaligned words 0.0016323129999999998
source sen 0.001616035
edge words 0.001606764
source syntax 0.0016057390000000001
network language 0.0016041649999999998
source span 0.001596932
affiliated source 0.001594861
neighbor words 0.001591809
baseline bleu 0.0015897580000000001
bleu baseline 0.0015897580000000001
network training 0.0015757969999999999
test sentence 0.0015727599999999999
large target 0.001532393
nnjm context 0.0014936329999999999
context vector 0.001485399
bilingual context 0.001474371
output layer 0.001474055
large context 0.0014732859999999999
translation 0.00147198
network models 0.001463256
probability models 0.0014614860000000001
same bleu 0.001448846
training time 0.0014335910000000001
feature score 0.001427242
layer matrix 0.001410597
phrase pairs 0.00139939
new language 0.001397766
input context 0.0013956419999999999
neural network 0.001395364
same sentence 0.001390844
nnjm feature 0.0013866299999999998
same feature 0.001382513
layer size 0.001381464
hidden layer 0.0013781309999999999
bleu improvement 0.001376112
target history 0.001368808
nist training 0.001365296
source 0.00135958
words 0.00135484
multiple target 0.001350501
parallel training 0.001317228
entire target 0.001312225
other features 0.00131116
decoding features 0.001309395
bleu improvements 0.001307484
context window 0.001304412
many language 0.001300309
language pairs 0.0012986949999999999
nnjm features 0.00129765
unaligned target 0.001294603
gual phrase 0.001290561
target hypothesis 0.001285451
phrase boundaries 0.001282184
bleu scores 0.001277294
layer scores 0.001269843
normal target 0.001267656
parallel sentence 0.001264912
baseline system 0.001256545
target neighbor 0.001254099
target vocabu 0.001253766
