semantic features 0.002634166
other features 0.0025012009999999998
role features 0.002297339
local features 0.002202691
test data 0.002184927
training data 0.002181536
various features 0.002124502
corresponding features 0.002092472
useful features 0.00208112
textual features 0.002074074
mantic features 0.002040694
trigram features 0.002039649
complementary features 0.002033891
noisy features 0.002029977
data set 0.00194523
task data 0.001894644
word insertions 0.001855648
following data 0.001839331
features 0.00177396
testing data 0.00174331
different feature 0.00173448
feature set 0.00167843
test corpus 0.001672207
newswire data 0.001670145
training corpus 0.001668816
real data 0.001668021
srl system 0.0016571630000000001
ldc data 0.001657106
test results 0.001643253
srl results 0.0016388470000000001
data sources 0.001626154
clec data 0.001619079
data sparseness 0.00160853
synthetic data 0.001607575
cial data 0.001607575
thetic data 0.001607575
language model 0.001605011
learning approach 0.001512569
feature vector 0.0015011780000000001
srl systems 0.0014859299999999999
learning algorithm 0.001474771
feature selection 0.0014534020000000001
semantic information 0.001450744
work srl 0.0014414739999999999
feature extraction 0.001420512
feature weight 0.001412413
distance feature 0.0014123360000000001
feature sets 0.0014106000000000001
corresponding feature 0.001403212
semantic role 0.0013835850000000001
sparse feature 0.0013833360000000002
training error 0.001354874
unrelated words 0.001317283
perceptron learning 0.00130083
semantic roles 0.001298037
other models 0.001291594
learning framework 0.001285738
domain test 0.0012763129999999998
local srl 0.001257752
srl local 0.001257752
similar performance 0.001256561
semantic fea 0.0012341309999999999
syntactic information 0.001233231
tron learning 0.001222655
current srl 0.001211639
second language 0.0012053419999999999
training examples 0.001197069
training corpora 0.001191527
training method 0.00118971
method srl 0.001188695
several srl 0.0011857
only srl 0.0011774749999999999
annotated training 0.0011674440000000001
other roles 0.001165072
test sets 0.001159327
newswire corpus 0.0011574250000000001
training process 0.001153949
srl output 0.001150479
level srl 0.001149107
training stage 0.001146546
semantic labels 0.001145929
semantic relations 0.00113424
experimental results 0.001132165
srl outputs 0.0011296119999999999
trigram language 0.001126438
performance differences 0.001125282
smt training 0.001123035
srl sys 0.001122005
smt system 0.001121141
semantic connection 0.0011162099999999999
syntactic structure 0.001102253
usage context 0.001101685
test dataset 0.0011008279999999999
significance test 0.001095892
lish corpus 0.00109576
training rounds 0.0010948400000000001
linear model 0.001091966
perceptron model 0.001090728
test purposes 0.001089328
training iterations 0.0010872780000000001
