word segmentation 0.00427137
chinese word 0.0035965470000000003
segmentation model 0.00319659
word segmenter 0.0031353600000000002
word seg 0.0031065240000000003
single word 0.00302044
word ctb 0.0030102080000000003
word segmen 0.0030013740000000002
meaningful word 0.0029854070000000003
word segmenta 0.0029567810000000003
nese word 0.002953854
model character 0.00287158
training data 0.0027156899999999998
segmentation knowledge 0.002407856
segmentation models 0.00232504
training corpus 0.002185449
training algorithm 0.0021306629999999997
segmentation standard 0.002115689
model accuracy 0.002105111
perceptron training 0.002088305
segmentation results 0.002072577
baseline model 0.002060033
time model 0.00205325
constraint segmentation 0.001999691
discriminative model 0.001942052
current model 0.0019325
classification model 0.001915113
model mapping 0.001870551
model mistakes 0.001868558
segmentation standards 0.001864041
additional training 0.0018439139999999999
training set 0.0017974669999999999
annotated data 0.001732402
standard words 0.001731559
ctb training 0.0017169479999999998
overall training 0.0016960109999999999
training pipeline 0.001692451
direct training 0.00167893
global training 0.001675975
training iterations 0.0016705499999999998
valuable training 0.001667553
training examples 0.0016611489999999998
criminative training 0.001651193
ditional training 0.0016498349999999999
raw data 0.001631174
character classifier 0.0016198089999999998
model 0.00161309
external data 0.0015859530000000002
segmentation 0.0015835
current character 0.0015779
character classification 0.001560513
tuation character 0.001555664
character classifi 0.00151421
kth character 0.00151421
feature vector 0.001477272
mented words 0.001462398
perceptron algorithm 0.001429748
learning algorithm 0.0014143749999999998
annotation information 0.0014119990000000002
local features 0.0014069500000000001
feature templates 0.001402282
new knowledge 0.0013981050000000002
training 0.00139461
perceptron learning 0.0013720170000000001
annotations different 0.0013406540000000002
ing information 0.001335962
complicated features 0.001332026
dependency parsing 0.001305505
chinese wikipedia 0.001304737
other characters 0.001276869
parsing performance 0.001273272
ing corpus 0.001266744
character 0.00125849
chinese treebank 0.00124065
chinese phrase 0.001239891
structural information 0.001234579
sentences figure 0.0012295869999999999
linguistic knowledge 0.001225579
ing models 0.001217445
other work 0.001202833
annotated corpus 0.001202161
words 0.00119937
penn chinese 0.001199304
different domains 0.001190038
knowledge source 0.001183017
baseline algorithm 0.001182996
annotated sentences 0.0011720559999999999
other sequence 0.0011706260000000001
chinese academy 0.001164861
different amounts 0.0011524080000000002
segmented sentences 0.001146748
different predications 0.001145971
different kind 0.001143647
information technology 0.001142929
baseline perceptron 0.001140638
different scales 0.001137425
information retrieval 0.001124348
information includ 0.001116918
whole corpus 0.001113787
valuable knowledge 0.001097299
