tree kernel 0.0043627100000000005
kernel model 0.003669
kernel features 0.0036275300000000003
other kernel 0.003156834
kernel method 0.003128911
kernel methods 0.003048825
new kernel 0.00299403
kernel models 0.002993469
linear kernel 0.002867703
top kernel 0.002859707
kernel fea 0.002826347
relevant kernel 0.002811532
fisher kernel 0.002799314
reranking kernel 0.002793642
efficient kernel 0.002783214
kernel meth 0.002726994
preference kernel 0.002720267
parse tree 0.002704015
structure tree 0.002497637
kernel 0.00244436
lexical tree 0.002336297
top tree 0.002333697
candidate tree 0.002282229
arbitrary tree 0.002256663
tree pair 0.002243003
ture tree 0.002219384
tree fragments 0.0021927410000000002
connected tree 0.002191082
tree yjk 0.002191082
probability model 0.0019977429999999997
feature vector 0.0018938079999999999
different model 0.0018653099999999998
many features 0.0017836650000000003
probabilistic model 0.0016857539999999998
feature vectors 0.00168538
model parameters 0.00164967
training data 0.0016250919999999999
input features 0.001612914
top kernels 0.001610457
feature extractor 0.0015945199999999999
sulting feature 0.001580829
tive model 0.001570359
model log 0.001555126
fisher kernels 0.001550064
trained model 0.0015311889999999999
kernels the 0.0015181390000000002
parse trees 0.001516997
exponential model 0.001512432
line model 0.0015047239999999998
ssn model 0.001500322
bilistic model 0.0014980009999999999
known kernels 0.00147082
underlying kernels 0.001470235
kernels tsuda 0.001468932
other trees 0.0014438060000000002
loss function 0.001345819
training algorithm 0.0013437520000000001
learning algorithm 0.0013381410000000002
discriminant function 0.001335738
standard training 0.001335593
exponential function 0.001329852
tic function 0.001322293
inant function 0.001316645
sigmoid function 0.001316645
partial parse 0.001313644
parsing method 0.0013121830000000002
structure trees 0.001310619
standard parse 0.001309086
same parsing 0.001308722
feature 0.00130693
parse history 0.001299245
network training 0.001292278
machine learning 0.0012860530000000001
other work 0.0012659
perceptron training 0.0012542830000000001
generative probability 0.001243083
classification task 0.001239376
new method 0.0012342210000000002
parsing methods 0.0012320970000000001
model 0.00122464
next word 0.0012242910000000002
large training 0.001222859
parse sequence 0.001199058
above probability 0.001198391
new classification 0.001196987
training sentences 0.0011952199999999999
kernels 0.00119511
different models 0.001189779
first vector 0.001189186
learning problems 0.001187257
features 0.00118317
output parse 0.001179134
parsing models 0.001176741
ing performance 0.001158339
learning bias 0.0011553520000000001
candidate parse 0.001149544
good performance 0.001149314
word pairs 0.001139077
same time 0.001138456
word vocabulary 0.001132636
