translation model 0.00555344
language translation 0.00527437
translation rule 0.005242709999999999
translation system 0.005169125
other translation 0.004830119999999999
many translation 0.004732109
possible translation 0.004729260999999999
parsing translation 0.004685045
translation models 0.004675889
translation systems 0.004654566
maximum translation 0.004628259
string translation 0.004584837
translation quality 0.0045776589999999995
translation train 0.004563561
translation stage 0.0045570229999999995
translation sacrifices 0.004514992999999999
translation metric 0.004512207999999999
translation derivations 0.004496015999999999
translation deriva 0.004488471
translation metrics 0.004486192999999999
translation yields 0.004486192999999999
translation 0.00420878
source source 0.00252
language model 0.00241025
source language 0.0023255899999999998
source tree 0.002235653
same model 0.002025316
source target 0.00199789
same source 0.001940656
word alignments 0.001929863
training data 0.0018704070000000001
model size 0.001862537
source side 0.0018494470000000002
syntactic parser 0.0018321890000000001
lation model 0.0018098699999999999
parallel data 0.001751291
same rules 0.001655876
data size 0.0016382470000000002
source trees 0.001634704
trigram model 0.0016248399999999998
syntactic structure 0.0016173519999999999
unique source 0.0015907030000000002
side tree 0.0015651
parse tree 0.0015574270000000001
source lan 0.001555186
parser probability 0.0014850520000000002
language text 0.001482976
maximum rule 0.001453409
language string 0.001441647
system score 0.001432206
single tree 0.001384361
rule extraction 0.001373018
structure parser 0.001370081
correct tree 0.0013587760000000001
rule height 0.001354814
string tree 0.00135171
tree string 0.00135171
model 0.00134466
string system 0.001336402
syntactic unit 0.00132729
baseline parser 0.001322907
same features 0.001298559
other words 0.001281838
source 0.00126
system objective 0.00125854
bleu score 0.001197366
correct parser 0.001175582
original parser 0.001161513
new feature 0.0011559600000000001
chinese parser 0.001145818
parser stage 0.001140702
other hypothesis 0.0011253729999999998
results table 0.001119841
ing performance 0.0011154379999999998
target string 0.0011139470000000001
parse forest 0.001104297
possible parse 0.001102255
parser errors 0.0011004930000000001
other information 0.001094463
charniak parser 0.001074917
mer training 0.001073263
parser gen 0.001073049
nese parser 0.001071777
standard algorithm 0.0010713419999999999
language 0.00106559
string figure 0.001058092
parallel text 0.001048307
main results 0.001047246
bilingual features 0.001046599
head words 0.001033155
previous work 0.001004798
different translations 9.9539E-4
single parse 9.90482E-4
feature weights 9.82443E-4
posterior probability 9.75744E-4
rules 9.7522E-4
system 9.60345E-4
ing accuracy 9.5958E-4
parse trees 9.56478E-4
recent work 9.54625E-4
