other model 0.003975726000000001
training data 0.00354629
model selection 0.003486243
total model 0.003477542
appropriate model 0.003460349
dynamic model 0.003405265
svmtool model 0.0033973980000000003
generalized model 0.003395149
alized model 0.003388195
model selec 0.003388195
eralized model 0.003388195
model 0.00313537
same data 0.00267103
ing data 0.002492566
heterogeneous data 0.002479922
testing data 0.002390551
external data 0.002342384
geneous data 0.002323531
same training 0.00209346
diverse training 0.0017899440000000001
training space 0.001770999
entire training 0.0017638800000000002
separate models 0.00165595
individual models 0.0016503050000000001
pos tag 0.001620012
generalized models 0.001571659
current word 0.001554111
feature set 0.001514704
training 0.00148436
pos tagging 0.001472677
pos tags 0.0013811869999999999
other systems 0.001355336
pos tagger 0.001355079
models 0.00131188
tagging accuracy 0.0012333230000000001
feature templates 0.001228324
other state 0.001223294
ing pos 0.001215748
different sources 0.001214479
different kinds 0.0012106649999999999
different genres 0.001209608
different doc 0.001197825
lexical features 0.001179144
other tasks 0.001158123
system tags 0.001148169
pos taggers 0.001144162
simple tagging 0.0011402650000000001
tagging approach 0.001112396
learning parameters 0.0011117129999999999
several dif 0.001098272
tagging algorithm 0.001090551
several rounds 0.0010672580000000002
tistical pos 0.001038164
same train 0.001027895
input sentences 0.0010101020000000001
first train 9.903120000000001E-4
combined tagging 9.89143E-4
bidirectional tagging 9.79473E-4
feature 9.67227E-4
tagging speed 9.6084E-4
cal features 9.43663E-4
tagging speeds 9.404039999999999E-4
journal corpus 9.40121E-4
input sentence 9.33392E-4
same source 9.32564E-4
future work 9.04317E-4
first time 8.8732E-4
comparable results 8.84746E-4
robust results 8.67004E-4
contemporary results 8.646979999999999E-4
accuracy comparisons 8.53272E-4
stanford tagger 8.49299E-4
related work 8.455909999999999E-4
stanford system 8.31426E-4
work toutanova 8.287489999999999E-4
ford tagger 8.23626E-4
general domain 8.1877E-4
alternative approach 8.18274E-4
broadcasting news 7.94396E-4
total number 7.82681E-4
evaluation sets 7.80105E-4
lexical items 7.796719999999999E-4
tion approach 7.78715E-4
character sequence 7.7856E-4
ghz machine 7.763570000000001E-4
selection approach 7.757040000000001E-4
individual file 7.71909E-4
robust performance 7.64065E-4
similarity distribution 7.64046E-4
alternative approaches 7.57445E-4
token experiments 7.5742E-4
ambiguity class 7.569650000000001E-4
sinorama news 7.54808E-4
cosine similarity 7.517540000000001E-4
unknown tokens 7.49052E-4
document frequency 7.376399999999999E-4
news magazine 7.32974E-4
simple algo 7.32951E-4
sequence classification 7.25567E-4
sinorama wsj 7.22525E-4
