training data 0.00219207
semantic role 0.00210963
semantic argument 0.0020156189999999997
test data 0.001861466
syntactic information 0.00182034
data wsj 0.0018125749999999999
syntactic head 0.0017473179999999999
syntactic path 0.001713539
different training 0.001665401
head word 0.001659048
data table 0.001587387
different feature 0.0015753899999999999
ing data 0.001569148
syntactic category 0.0015574679999999998
syntactic infor 0.001519424
testing data 0.001508943
large training 0.001476097
data sets 0.001466705
experimental data 0.001464802
feature set 0.001456498
training sentences 0.001453157
important role 0.0014519659999999998
learning algorithm 0.001443955
word group 0.00141901
arguments system 0.0014148329999999999
output training 0.001393061
second argument 0.0013928550000000001
brown corpus 0.001389399
training dataset 0.001388312
base role 0.001387259
system accuracy 0.001385517
role classification 0.0013750939999999999
predicate argument 0.001369291
parser 0.00134025
training times 0.001336317
learning techniques 0.001327295
role labeling 0.001324074
feature representation 0.001322817
training datasets 0.0013175919999999998
pml training 0.0012964069999999999
machine learning 0.001295812
training sets 0.001294235
argument classification 0.0012810830000000001
combined training 0.001279456
role classifier 0.001274547
corpus test 0.001268059
work features 0.0012676599999999999
training instances 0.001267064
argument distance 0.001260866
para argument 0.001254187
other systems 0.001252297
suitable role 0.001247834
known argument 0.001241335
argument recognition 0.001234832
basic features 0.0012170240000000001
learning algorithms 0.001213257
system train 0.00120787
argument identification 0.0012059710000000001
full feature 0.001204718
feature representations 0.001203989
pml system 0.001203746
feature subsets 0.001189409
language domain 0.0011858279999999999
learning curve 0.001180502
feature weighting 0.001177523
preposition feature 0.001174641
system architecture 0.001172087
learning architecture 0.001167295
feature dimension 0.001165407
feature reduc 0.001165407
different systems 0.001164445
tive learning 0.001158765
snow learning 0.001158765
lazy learning 0.001158765
argument bounda 0.001152756
model 0.00113439
known features 0.001128104
different distributions 0.0011063029999999999
different results 0.0010933
brown cor 0.00109321
different srl 0.001090849
features most 0.001061144
parse trees 0.001048123
brown sent 0.001045056
parse tree 0.001043329
test result 0.001036847
training 0.0010098
parse constituent 0.001007251
phrase path 9.978460000000001E-4
other modifications 9.95977E-4
wsj section 9.77986E-4
probability distribution 9.70335E-4
train test 9.69927E-4
test tasks 9.658819999999999E-4
performance drops 9.651870000000001E-4
acceptable performance 9.651870000000001E-4
different methods 9.64503E-4
test sets 9.63631E-4
first row 9.63195E-4
basic algorithm 9.536740000000001E-4
