other features 0.003411167
context features 0.0031353870000000003
adjective features 0.003011194
lex features 0.0029880690000000003
count features 0.0029468320000000004
binary features 0.0028986930000000004
collocation features 0.0028853510000000004
indicator features 0.0028445140000000002
bilexical features 0.00282913
ble features 0.002828091
clude features 0.002818624
suffix features 0.0028151170000000002
features lapata 0.0028151170000000002
features 0.00256687
training data 0.0025323100000000003
test data 0.002376094
new data 0.002112857
ing data 0.001982594
standard data 0.001936136
model results 0.001913072
labeled data 0.00190397
source data 0.001903207
unlabeled data 0.00188269
medline data 0.0018747570000000001
training corpus 0.001871973
data sets 0.001807831
data the 0.0018016870000000002
auxiliary data 0.001780968
data sparsity 0.001770834
representative data 0.001766158
data sizes 0.0017652940000000002
ood data 0.001764323
training domain 0.0017391120000000001
web model 0.001724645
domain test 0.0015828959999999999
word form 0.001568547
dependency model 0.0015244599999999998
feature value 0.0015239939999999999
corpus counts 0.001519196
web corpus 0.001508348
training set 0.00150054
word pairs 0.001490827
multiple feature 0.001487495
single model 0.001463828
particular word 0.001422399
different domain 0.0014189329999999998
large corpus 0.0014134220000000001
feature label 0.001411902
word snapshot 0.001402897
labeled training 0.00140218
confusable word 0.001389253
feature vectors 0.0013883279999999999
fusable word 0.001383365
likely word 0.001383365
feature space 0.0013832739999999999
training pairs 0.001371737
feature weights 0.001368009
feature classes 0.001364302
training corpora 0.001357755
indicator feature 0.001349984
little training 0.0013495970000000001
training examples 0.001348712
test set 0.001344324
training vectors 0.0013312480000000002
feature views 0.0013278
adjacency model 0.0013274569999999998
dency model 0.0013239209999999998
dicator feature 0.001322059
economical model 0.001321933
cency model 0.001321933
pendency model 0.001321933
new domain 0.001319659
web counts 0.001314118
nyt training 0.001311516
training size 0.001292363
other tasks 0.0012733970000000001
training exam 0.001265142
training sizes 0.001263504
training collections 0.0012623410000000002
context counts 0.0012309999999999999
good performance 0.001228253
match words 0.001223235
correction results 0.001220549
test pairs 0.001215521
test sentence 0.001199206
test examples 0.001192496
lex system 0.001184224
class baseline 0.0011781019999999999
confusable words 0.0011718499999999999
mixed words 0.00116604
web query 0.001163744
unsupervised system 0.001163233
bracketing results 0.0011603960000000002
national corpus 0.001155784
domain tasks 0.001152952
test sets 0.001149825
gutenberg corpus 0.001148715
standard domain 0.001142938
disambiguation results 0.001140739
domain usage 0.001131919
