coreference data 0.0030853900000000004
data set 0.003069459
training data 0.002978695
different data 0.002815377
labeled data 0.002703492
unlabeled data 0.0026997880000000003
data accuracy 0.002614191
data pool 0.0025291230000000003
nlp data 0.0024756960000000003
data sets 0.0024568330000000003
entire data 0.0024497440000000002
artificial data 0.002433128
data redundancy 0.002426614
tial data 0.002426614
recursive data 0.002426614
feature set 0.002161329
learning algorithm 0.002146559
coreference classifier 0.0021281900000000003
bootstrapping algorithm 0.002000909
different learning 0.001921287
other classifier 0.0018940650000000001
learning algorithms 0.001819586
training set 0.0017647140000000001
training algorithm 0.001735904
natural feature 0.0017286649999999999
supervised learning 0.0017118300000000001
feature split 0.001695064
bootstrapping algorithms 0.001673936
selection method 0.001642601
class value 0.001629555
feature values 0.0016258549999999998
same set 0.0016239100000000001
classifier prediction 0.001610777
performance deterioration 0.001598048
semantic class 0.0015976800000000002
full feature 0.001584018
ent learning 0.001582003
feature type 0.0015749689999999998
view bootstrapping 0.001568811
performance improvements 0.001562377
bootstrapping process 0.00155642
active learning 0.001552833
independent feature 0.0015501
learning curve 0.00154486
able feature 0.0015431639999999999
binary classifier 0.001536449
bootstrapping experiments 0.001536396
performance gains 0.001533789
feature subsets 0.0015337279999999998
feature val 0.001533551
labeled instances 0.0015205890000000001
disjoint feature 0.001519915
obvious feature 0.001519915
unlabeled instances 0.001516885
single classifier 0.00150247
coreference classifiers 0.001502247
similar performance 0.001500753
clustering algorithm 0.001499897
class distribution 0.001480577
classifier partition 0.0014741110000000002
classifier highest 0.001473216
new instances 0.001471072
negative class 0.001465275
coreference system 0.001462179
performance trend 0.001461613
ranking instances 0.0014577180000000002
performance trends 0.001452174
marginal performance 0.001452174
performance guaran 0.001452174
performance dete 0.001452174
superior performance 0.001452174
yield performance 0.001452174
performance deterio 0.001452174
provable performance 0.001452174
new information 0.001419161
coreference task 0.001416985
coreference resolution 0.001412834
training algorithms 0.001408931
bootstrapping basis 0.001403938
bootstrapping procedure 0.001401869
bootstrapping curve 0.0013992100000000001
bootstrapping schemes 0.001387396
ours bootstrapping 0.001387396
ing instances 0.001377628
same different 0.001369828
new method 0.0013688799999999998
supervised algorithm 0.001363129
class boundaries 0.001358843
method figure 0.001356459
ranking method 0.0013555260000000001
different different 0.001347314
equivalence class 0.001332111
standard coreference 0.001330589
class distributions 0.001330347
same label 0.001329887
skewed class 0.0013236670000000002
ing algorithm 0.00131774
class priors 0.001315868
class densities 0.001313149
minority class 0.001311389
