translation model 0.0025639
language model 0.0022128070000000002
model score 0.001934538
training set 0.001771063
training time 0.0017602339999999998
translation rule 0.001749679
large training 0.001737186
model fixes 0.0017320640000000002
model tai 0.0017320640000000002
same translation 0.0017239
partial translation 0.001708457
discriminative training 0.001672902
training corpus 0.0016307119999999999
training algorithms 0.001566927
perceptron training 0.001552164
derivation tree 0.001545175
training iteration 0.001536755
ous training 0.0015180810000000001
training algo 0.001515353
maxforce training 0.0015075169999999999
ious training 0.00150415
decoding hypergraph 0.0014794489999999999
translation probabilities 0.001469907
model 0.00146563
machine translation 0.001456755
hiero translation 0.001410423
reference translation 0.00140964
lexical translation 0.0014052259999999999
decoding process 0.001393043
english word 0.001387926
translation hypergraphs 0.0013870039999999998
direct translation 0.0013768819999999998
translation prob 0.0013650849999999998
decoding reachability 0.001326945
gold tree 0.001324745
whole decoding 0.001316198
hiero decoding 0.001302677
features hypergraph 0.0012884099999999998
real decoding 0.001282453
other words 0.001279717
forced decoding 0.001266747
training 0.00123708
other tuning 0.0012326239999999999
sentence length 0.001224223
rule set 0.001185392
partial derivation 0.0011635130000000001
chinese words 0.0011588100000000001
sparse features 0.001138744
bleu improvement 0.001136261
full rule 0.00113344
gram language 0.0011272040000000001
partial derivations 0.0011263990000000002
features design 0.001116871
tuning methods 0.0011162989999999999
bleu scores 0.001116185
many derivations 0.0011137170000000002
input sentence 0.001110465
feature vector 0.001099768
natural language 0.001099681
translation 0.00109827
lexicalized features 0.001089833
dense features 0.001075632
word clus 0.001067901
word types 0.001053733
boundary word 0.001045262
language processing 0.001043289
bleu points 0.001042379
full derivation 0.001035357
variable hypergraph 0.001033535
nese word 0.001029796
rule extraction 0.001008468
source sentence 9.95858E-4
derivation figure 9.94338E-4
penalty weight 9.92281E-4
decoding 9.90524E-4
other hand 9.88477E-4
such derivations 9.86822E-4
following experiments 9.84113E-4
small tuning 9.76967E-4
sentence pairs 9.76762E-4
test set 9.72553E-4
length penalty 9.66764E-4
viterbi derivation 9.65102E-4
short sentences 9.55745E-4
sentence pair 9.557389999999999E-4
first time 9.507109999999999E-4
hiero rules 9.50398E-4
loose rule 9.429880000000001E-4
partial deriva 9.37597E-4
same reference 9.369999999999999E-4
cdec system 9.31638E-4
ing corpus 9.30217E-4
test hypergraph 9.27495E-4
glue rules 9.250389999999999E-4
synchronous grammar 9.21223E-4
work figure 9.130480000000001E-4
reference words 9.098929999999999E-4
violation method 9.096429999999999E-4
deductive system 8.96886E-4
great results 8.87781E-4
