model performance 0.003676574
language model 0.003586997
model test 0.003582136
training data 0.00354493
prior model 0.003418488
memm model 0.003398724
probability model 0.00339646
model size 0.0033299370000000003
model results 0.003304366
entropy model 0.003294031
model adaptation 0.003256599
capitalization model 0.003245983
markov model 0.003214142
test data 0.003214086
probabilistic model 0.003206916
background model 0.0031865590000000003
new model 0.003185386
tion model 0.0031710840000000002
wsj model 0.0031706710000000003
model parameters 0.003161264
transition model 0.0031111750000000003
model sizes 0.003100592
model param 0.0031000470000000003
ability model 0.003097036
ent model 0.0030937960000000002
exponential model 0.003091345
ground model 0.003087392
model adapt 0.003086704
data size 0.0029618870000000003
news data 0.0029332530000000002
data sets 0.0028944580000000004
adaptation data 0.0028885490000000002
specific data 0.002878699
model 0.00283407
background data 0.0028185090000000003
development data 0.0028015780000000003
data cap 0.002798576
data partition 0.0027760140000000003
primetime data 0.002775557
right data 0.0027546130000000004
abc data 0.0027479690000000003
ground data 0.002719342
adapt data 0.002718654
opment data 0.002717619
word features 0.002189199
feature set 0.002032837
feature weights 0.001937081
feature sets 0.001902588
adaptation feature 0.001896679
word sequence 0.0018597510000000002
feature selection 0.001785151
fadapt feature 0.001764343
fbackground feature 0.001761441
first word 0.001696169
current word 0.0016400170000000001
models performance 0.001639748
training set 0.001637597
microsoft word 0.001533283
lowercase word 0.001533283
word processors 0.001533283
modern word 0.001533283
tag sequence 0.001519964
training procedure 0.001505529
adaptation training 0.001501439
feature 0.00147415
discriminative training 0.0014652929999999999
divergence training 0.001460886
background training 0.0014313989999999999
set performance 0.001401191
maxent models 0.001391565
possible tag 0.001368355
memm models 0.001361898
probability models 0.001359634
training logl 0.001336419
ground training 0.001332232
mindiv training 0.001330469
test set 0.001306753
models size 0.001293111
approach models 0.001280005
other text 0.0012688360000000002
few features 0.001265542
adaptation performance 0.001265033
background features 0.001261688
new features 0.001260515
entropy models 0.001257205
relative performance 0.001250198
performance improvement 0.00123916
case information 0.0012376969999999998
likely tag 0.001234719
tag assignments 0.001233529
test time 0.00121509
contextual features 0.001200428
features fadapt 0.001199392
frequent tag 0.0011977189999999999
such words 0.001196539
arate tag 0.001193011
markov models 0.001177316
test sets 0.001176504
maxent probability 0.001156711
background models 0.001149733
