language model 0.001836346
order model 0.001737697
training data 0.001661315
gram model 0.001599685
model building 0.001547715
first word 0.0014627569999999999
word sense 0.00144189
word vocabulary 0.001354924
model 0.00134066
word frequencies 0.001331624
word cor 0.001322893
word brown 0.001314313
times word 0.0013065639999999999
last word 0.001301099
word sequences 0.001293025
data sets 0.001272585
count words 0.0012556149999999999
language models 0.001248373
smoothing algorithm 0.001186529
training corpus 0.001088177
frequency counts 0.001067926
new algorithm 0.001061962
complete counts 0.0010421480000000001
necessary counts 0.001029826
extra counts 0.00102825
smoothing method 0.001021486
interpolated models 0.001003121
missing counts 9.96783E-4
tra counts 9.9519E-4
plainn counts 9.92381E-4
unique words 9.838E-4
distinct words 9.73186E-4
machine translation 9.72786E-4
unseen words 9.67201E-4
ing method 9.66817E-4
lated models 9.62004E-4
different methods 9.2633E-4
previous language 9.035029999999999E-4
new method 8.969189999999999E-4
other parameters 8.91649E-4
standard smoothing 8.748720000000001E-4
language modeling 8.47812E-4
such datasets 8.41313E-4
probability estimates 8.330950000000001E-4
method all 8.322379999999999E-4
such redistribution 8.27443E-4
web sentences 8.18069E-4
discount form 8.170699999999999E-4
dirichlet form 8.124269999999999E-4
low count 8.11133E-4
order distribution 8.023220000000001E-4
new smoothing 8.02175E-4
smoothing methods 7.852040000000001E-4
counts 7.84333E-4
total count 7.74328E-4
interpolated form 7.63596E-4
prior form 7.592699999999999E-4
gram language 7.547109999999999E-4
models 7.52687E-4
same idea 7.4646E-4
sentence tags 7.38828E-4
bayesian approach 7.374230000000001E-4
words 7.34537E-4
various smoothing 7.326590000000001E-4
count abc 7.3142E-4
ing methods 7.30535E-4
nonzero count 7.28915E-4
large corpora 7.28866E-4
algorithm 7.23158E-4
form mackay 7.21764E-4
reasonable form 7.21764E-4
good results 6.96833E-4
training 6.90977E-4
independent probability 6.89243E-4
probability mass 6.77673E-4
smoothing meth 6.743750000000001E-4
order distributions 6.711709999999999E-4
large datasets 6.69797E-4
parameter optimization 6.55935E-4
case range 6.45191E-4
lexical substitution 6.34018E-4
brown corpus 6.322330000000001E-4
naive application 6.15484E-4
single idea 6.14202E-4
full potential 6.13402E-4
absolute discounting 6.09489E-4
est order 6.0851E-4
guage modeling 6.022930000000001E-4
general structure 6.00436E-4
modified estimates 5.99369E-4
right hand 5.95607E-4
perplexity reduction 5.94612E-4
original study 5.88697E-4
fixed constant 5.86315E-4
sense disambigua 5.798180000000001E-4
low frequency 5.73648E-4
absolute discount 5.73521E-4
modeling stud 5.608830000000001E-4
modeling tools 5.608830000000001E-4
sonable alternative 5.59555E-4
