possible features 0.002545492
lexical features 0.002543998
word error 0.002540606
word context 0.002517745
features first 0.002493195
dictionary features 0.00242733
word sequence 0.002402338
unigram features 0.002365087
effective features 0.002354091
possible word 0.002334182
bigram features 0.002333141
features designing 0.002323278
ical features 0.002313326
erate features 0.002311073
feature selection 0.002236014
single word 0.002235285
word sequences 0.00216755
various feature 0.002165139
tal feature 0.002147617
feature tem 0.0021473539999999998
feature functions 0.002142103
word transcriptions 0.002141648
bigram feature 0.0021348310000000002
feature templates 0.002134447
manual word 0.0021273810000000002
feature template 0.002116836
cremental feature 0.002113344
features 0.00207969
several words 0.001929325
language model 0.001899815
feature 0.00188138
ing words 0.0017951129999999999
rare words 0.001778005
vocabulary words 0.0017442579999999998
consecutive words 0.0017353969999999999
map words 0.0017225489999999999
maxent model 0.001687159
surrounding words 0.001684105
model maximum 0.0016752709999999999
speech data 0.001669948
entropy model 0.001665967
training data 0.0016610029999999999
test data 0.0016009050000000001
background model 0.0015984609999999998
ent model 0.001554048
tropy model 0.001527487
model focus 0.001520704
number information 0.001455522
words 0.00145113
ing data 0.001355343
available data 0.001352068
speech text 0.001337118
data likelihood 0.0013197249999999999
model 0.00128967
information extraction 0.001278616
data sparsity 0.001258025
different systems 0.001244814
test set 0.0012217719999999999
ing information 0.001218678
large number 0.001196796
recognition system 0.001193209
baseline system 0.0011909619999999998
speech recognition 0.001190916
particular tag 0.001177071
caller information 0.001168684
key information 0.001146813
tag contexts 0.001130711
information bearing 0.0011238189999999999
different times 0.001121033
information extract 0.00111907
true information 0.001112062
tag sequences 0.001109388
different sys 0.001075356
evaluation test 0.001074549
first system 0.001074386
second system 0.0010730100000000001
different transitions 0.001061508
tagging problem 0.001061092
error rate 0.001034329
training procedure 9.98026E-4
same way 9.8429E-4
number structure 9.82176E-4
string error 9.79052E-4
human tagger 9.778270000000001E-4
recognition work 9.75265E-4
text sources 9.717840000000001E-4
number trigger 9.702980000000001E-4
phone number 9.66967E-4
training examples 9.61333E-4
other hand 9.56855E-4
speech sources 9.51842E-4
related speech 9.469999999999999E-4
section speech 9.45365E-4
labeled training 9.452E-4
third system 9.382500000000001E-4
flex system 9.31674E-4
number dictionary 9.284670000000001E-4
background language 9.18936E-4
statistical models 9.15021E-4
speech recognizer 9.149869999999999E-4
