language model 0.00220758
training data 0.002110778
different training 0.0019789480000000003
training corpus 0.001894192
data set 0.001816063
set data 0.001816063
language models 0.0017723819999999999
evaluation data 0.0017245519999999999
topic word 0.0017066019999999998
word topic 0.0017066019999999998
training set 0.0016849410000000001
other model 0.001662481
baseline language 0.001622916
topic model 0.0016215919999999998
feature model 0.001581945
topic words 0.0015205779999999999
word error 0.00151779
interpolation model 0.001500645
domain language 0.00148812
natural language 0.001475764
ing data 0.001470417
language modeling 0.001467453
available data 0.001457183
language applications 0.001451909
full training 0.0014447490000000002
training corpora 0.0014382980000000002
data sets 0.001431679
data segmentation 0.001427956
development data 0.0014211999999999999
model perplexity 0.0014156099999999999
wer model 0.001408664
same features 0.00140734
many training 0.001399868
target data 0.0013837709999999998
test set 0.0013760600000000001
absolute word 0.00137338
different combination 0.001372061
various training 0.0013675990000000002
mismatched data 0.0013630699999999999
data mismatch 0.001358509
target word 0.001357951
model parameters 0.001354386
model parameter 0.00135149
opment data 0.0013496979999999999
audio data 0.0013487369999999998
segment word 0.0013327629999999998
model combinations 0.001295937
matched training 0.0012870190000000001
model dev 0.001260962
acoustic model 0.001259048
training documents 0.001252651
particular corpus 0.001240134
model spe 0.001238451
textbook corpus 0.001227983
different parti 0.00122595
other feature 0.001224186
such features 0.00121844
training doc 0.001217687
training conditions 0.001217687
probability features 0.0012101350000000002
language 0.00119746
corpus com 0.001196641
example corpus 0.001192977
topic probability 0.00117008
metadata information 0.001163529
same wer 0.001154357
lectures corpus 0.0011535719999999998
switchboard corpus 0.0011525239999999998
corpus docu 0.001133189
probability feature 0.0011304330000000001
segmentation information 0.001117558
weighting features 0.001113194
set performance 0.001109022
error test 0.001103607
set perplexity 0.001100603
set values 0.0010977180000000001
baseline features 0.001076983
features perplexity 0.001057017
evaluation corpora 0.001052072
document topic 0.001036347
various features 0.001029298
high probability 0.0010235980000000001
such corpora 0.0010153830000000001
model 0.00101012
development set 9.95363E-4
entropy features 9.92789E-4
test sets 9.91676E-4
set size 9.91016E-4
training 9.89828E-4
topic distribution 9.84721E-4
other approaches 9.77576E-4
feature values 9.7443E-4
absolute test 9.59197E-4
set documents 9.579359999999999E-4
feature combination 9.547659999999999E-4
various models 9.526929999999999E-4
various feature 9.49596E-4
feature selection 9.49071E-4
individual models 9.352329999999999E-4
dev test 9.317889999999999E-4
