training model 0.003074619
tagging model 0.002735652
model parameters 0.002732016
uniform model 0.002592814
initial model 0.002589878
induction model 0.002523832
conditional model 0.002518702
single model 0.002510135
tion model 0.002494988
corresponding model 0.002483807
model events 0.002476943
final model 0.002462501
model event 0.002425451
model let 0.002414961
channel model 0.002412277
complex model 0.002408871
model deficiency 0.002398379
model 0.00215949
pos tag 0.0019782619999999997
tag dictionary 0.0017970199999999999
training data 0.001704421
such models 0.0016735080000000002
training word 0.001648531
tag sequence 0.001625935
tag sequences 0.0016246849999999998
particular tag 0.001620907
other training 0.001619621
same feature 0.001606625
tag tokens 0.0015871689999999998
different training 0.001584886
tag symbols 0.0015759939999999998
tag yield 0.00153418
same data 0.001532224
tag strings 0.001528768
entropy models 0.001522215
training likelihood 0.001475413
training corpus 0.0014683769999999999
nlp models 0.0014517940000000002
training tags 0.001431643
supervised models 0.001400435
training set 0.001383755
markov models 0.001379314
objective function 0.001321243
clustering models 0.001320016
ccm models 0.001312845
deficient models 0.001297692
language data 0.001295802
likelihood function 0.001294808
complex models 0.001293701
tic models 0.001289334
unsupervised data 0.001276554
supervised training 0.001271244
test data 0.0012625330000000001
pos tagging 0.0012593539999999999
unlabeled training 0.001247446
accuracy results 0.001242642
final training 0.00121814
test accuracy 0.0012052970000000001
set accuracy 0.001200682
pos tags 0.001199706
local maximum 0.001194632
ing learning 0.001181128
feature counts 0.001180256
training sets 0.001179058
correct training 0.00117661
training cross 0.001171338
learning method 0.0011670449999999998
training procedures 0.001150407
criminative training 0.001149907
respective training 0.001148448
other words 0.0011459690000000002
parameter estimation 0.001144085
hidden data 0.0011362730000000001
unlabeled data 0.001121609
new algorithm 0.001107212
yield feature 0.001102803
accuracy gain 0.001085623
accuracy improvement 0.001084998
same criterion 0.001084327
tagging dictionary 0.001078112
statistical grammar 0.0010693299999999998
clustering data 0.001064988
word sequence 0.001064267
other languages 0.001058925
unsupervised parameter 0.001058352
other term 0.001056754
unsupervised learning 0.001052317
labeled data 0.001052005
models 0.00104432
tive function 0.001042239
other hand 0.0010365630000000001
new parameters 0.001033977
test likelihood 0.001033525
test corpus 0.001026489
grammar induction 0.0010229829999999999
concave function 0.001017478
local optima 0.0010161789999999999
expected accuracy 0.001014752
local maxima 0.001013492
local max 0.00101289
