chinese word 0.0027478
chinese srl 0.00263682
predicate word 0.002466117
syntactic information 0.002362931
frame features 0.002347257
first word 0.002340252
srl performance 0.002298071
new features 0.002294495
baseline features 0.002276016
many features 0.002266076
syntactic parse 0.002251637
formation features 0.002239704
following features 0.002201653
tion features 0.002172824
head word 0.00216664
srl system 0.002151801
above features 0.002149215
tional features 0.002146955
additional features 0.002146695
features yield 0.00213272
srl results 0.0021309650000000003
features com 0.002125398
type word 0.0021188830000000002
features most 0.002114204
srl problem 0.0020713990000000002
syntactic structure 0.002042551
word position 0.002027867
automatic word 0.002026514
word fea 0.002026172
tag feature 0.002019682
semantic chunk 0.002011243
semantic role 0.001993777
word formation 0.001983134
syntactic trees 0.001978526
srl research 0.0019772780000000003
word segmentation 0.001947012
word level 0.001938204
chinese words 0.0019257809999999999
word length 0.001925544
english srl 0.001922131
previous srl 0.0019212230000000001
collocation word 0.001916318
semantic arguments 0.001908112
only word 0.001898268
trivial word 0.001896513
frame feature 0.0018954269999999999
srl models 0.0018868090000000001
srl systems 0.0018843730000000002
word constraints 0.001872271
srl methods 0.0018629240000000002
coarse word 0.001862549
features 0.00185914
cate word 0.001859014
word formations 0.00185638
precision srl 0.0018307660000000002
comparison srl 0.00182463
architecture srl 0.001820495
syntactic parses 0.001808055
semantic type 0.001807773
independent srl 0.0018069030000000001
recall srl 0.0017933040000000001
semantic labels 0.0017909609999999998
srl tasks 0.001788918
syntactic chunks 0.001786037
overall srl 0.0017787610000000002
syntactic chunking 0.001757429
whole srl 0.001757122
semantic classification 0.001756339
encouraging srl 0.0017541840000000002
srl sys 0.0017509320000000002
porate srl 0.001744818
glish srl 0.001744818
feature templates 0.0017387
argument labels 0.001709511
semantic roles 0.0016901389999999998
semantic chunks 0.001645737
argument position 0.0016353070000000001
chinese language 0.0016346919999999999
semantic chunking 0.001617129
semantic types 0.001602964
training data 0.001575354
automatic chinese 0.001569174
argument identification 0.001559333
chinese verbs 0.0015550989999999999
core argument 0.00154098
chinese propbank 0.0015272519999999998
chinese src 0.001524644
argument boundaries 0.001517843
parsing performance 0.001505915
different chunk 0.001477316
whole argument 0.001475542
chinese chunking 0.001470899
system performance 0.001466692
chinese seman 0.001461464
first chunk 0.001457465
chinese nouns 0.0014342729999999998
syntactic 0.00143176
relevant information 0.001420868
chinese penn 0.0014167519999999999
chinese char 0.001414212
