chinese word 0.00313824
language model 0.0029822
pinyin model 0.0029695499999999996
model pinyin 0.0029695499999999996
english model 0.00288837
english word 0.00286468
chinese language 0.00286185
input data 0.00268994
chinese input 0.00256786
english error 0.00247795
spelling model 0.002454063
chinese character 0.002410607
chinese words 0.002385394
pinyin input 0.00237548
same data 0.0023735320000000002
training data 0.002367237
chinese spelling 0.002333713
input error 0.00227779
word sequence 0.0022487930000000002
english words 0.002111834
statistical data 0.002092835
chinese characters 0.002087205
word string 0.002084535
syllable model 0.002060278
english spelling 0.002060153
typing data 0.002054032
typing model 0.0020523019999999998
model weight 0.00205025
markov model 0.002046531
real data 0.002040569
testing data 0.001976631
word end 0.001971916
trigram model 0.0019566979999999998
model parameters 0.001943233
word sequences 0.0019413020000000002
trigram word 0.0019330080000000002
typing chinese 0.001931952
chinese users 0.001930631
word perplexity 0.0019248730000000001
model base 0.0019200929999999999
data requirement 0.0019076660000000001
acoustic model 0.001907003
system errors 0.001896539
word slm 0.001883431
chinese dialect 0.001879113
input method 0.001859753
correct pinyin 0.001846543
text input 0.001831512
english characters 0.001813645
mandarin chinese 0.0017934
statistical language 0.001791025
probable chinese 0.0017891979999999999
chinese document 0.001786956
chinese slm 0.001786771
chinese cstar 0.001785257
pinyin typing 0.001739572
typing pinyin 0.001739572
user input 0.001735468
type english 0.00167927
input sequence 0.0016784130000000001
language mode 0.001674563
rate english 0.001673613
letter english 0.001672003
many system 0.001669744
syllable english 0.001666368
small error 0.0016620860000000001
language texts 0.001657421
error rate 0.001657103
trigram language 0.0016566180000000001
real english 0.001644929
model 0.00164114
english letters 0.001638609
english models 0.0016304919999999999
pinyin strings 0.0016261819999999999
english syllables 0.00161671
mixed language 0.001610969
legal pinyin 0.0016078359999999999
language modelling 0.001605396
powerful language 0.001605396
modeless pinyin 0.001603025
conversion errors 0.0015996650000000001
further error 0.001570211
correct input 0.001565203
baseline system 0.001562638
total error 0.001555791
input key 0.001547039
spelling correction 0.001542158
input methods 0.001528347
english tri 0.001523458
character sequence 0.0015211600000000001
chinese 0.00152079
error reduction 0.00151996
training corpus 0.001519482
unseen english 0.001518097
triletter english 0.001513169
system performance 0.001510073
error rates 0.001507848
error tolerance 0.001496092
error increase 0.001495836
real system 0.001485289
