training data 0.00337903
test data 0.0028177329999999998
same data 0.002436285
data sequence 0.002398123
ing data 0.00235618
data likelihood 0.002247405
testing data 0.002236052
training corpus 0.002220731
additional data 0.00221388
data preprocessing 0.0021895589999999998
data scales 0.0021895589999999998
training set 0.002179904
maxent model 0.002000176
based model 0.00190346
our model 0.001864197
model portion 0.001852634
model brown 0.001842768
training process 0.001833945
relevant features 0.001814487
reranking model 0.001814018
ent model 0.0018047200000000001
original training 0.0017923939999999999
comprehensive model 0.001789746
feature vector 0.001773105
numeric features 0.001754432
training time 0.001745393
arbitrary features 0.001724845
tor features 0.001713511
binary feature 0.001690341
test corpus 0.001659434
feature selection 0.001622349
test set 0.001618607
separate feature 0.001583945
model 0.00153652
features 0.00145961
training 0.00144112
same class 0.001410037
test corpora 0.001339934
specific class 0.001333772
class based 0.001278602
feature 0.00126589
test sets 0.001251084
original test 0.001231097
other cor 0.001225418
wsj test 0.00121009
clustering method 0.0012086570000000001
word type 0.00120517
class ordering 0.001203183
preferences class 0.0011905890000000001
word types 0.001189834
mitchell class 0.00118949
test time 0.001184096
corpus number 0.0011832260000000001
other permutations 0.001180265
transitivity method 0.001177155
other entries 0.001171488
nanc class 0.001171034
other possibilities 0.0011674300000000001
system performance 0.0011458430000000001
switchboard test 0.001142323
use corpus 0.001140029
test run 0.001131649
possible modifier 0.0011284020000000001
wsj corpus 0.001109878
modifier order 0.00108727
brown corpus 0.001085859
corpus token 0.0010810070000000001
newswire corpus 0.00107587
type accuracy 0.001069398
prediction accuracy 0.001051572
final set 0.001049461
learning curves 0.001042594
switchboard corpus 0.001042111
nanc corpus 0.001038983
ent set 0.001006984
modifier sequence 0.001005683
ing problem 0.0010049640000000001
different parts 9.97077E-4
entire set 9.96514E-4
unordered set 9.91327E-4
token accuracy 9.893459999999999E-4
noun phrase 9.85892E-4
maxent approach 9.79565E-4
head noun 9.70942E-4
computational approaches 9.64606E-4
maximum error 9.601309999999999E-4
noun phrases 9.41451E-4
semantic property 9.316750000000001E-4
noun andm 9.311510000000001E-4
common modifier 9.301380000000001E-4
method 9.22939E-4
pos tag 9.19857E-4
weight vector 9.13463E-4
class 9.11662E-4
original system 9.115E-4
modifier sequences 9.065330000000001E-4
order modifiers 8.982910000000001E-4
indicator function 8.982759999999999E-4
certain modifier 8.98086E-4
based approach 8.82849E-4
