language model 0.0026887400000000002
language models 0.0024132199999999998
randomised data 0.001957859
baseline model 0.0019467619999999999
word corpus 0.001917648
data scores 0.001915964
data structure 0.0018789050000000002
lossless data 0.001872715
parallel data 0.001850901
data structures 0.001841662
europarl data 0.001811816
associative data 0.0017994530000000001
model tests 0.001793068
eurpoparl data 0.001783805
trigram model 0.001778126
additional language 0.001694481
filter models 0.0016552379999999998
baseline language 0.0016076620000000002
filter language 0.0015916580000000001
entropy models 0.001575358
target language 0.00154801
natural language 0.001517686
model 0.00151392
trigram models 0.001502606
comparison models 0.001498127
word frequencies 0.001489965
feature function 0.001483862
gram language 0.00146226
training corpus 0.001436435
language cor 0.001430375
language modelling 0.001423217
error rate 0.001350623
translation performance 0.001336373
machine translation 0.001327089
hash values 0.001307224
error frequency 0.001304386
translation experiments 0.001301962
rate training 0.001295202
test frequency 0.0012718410000000001
translation process 0.001267342
training frequency 0.0012489649999999999
different approach 0.001242611
models 0.0012384
test time 0.001222455
additional space 0.0012192560000000002
bloom error 0.0012181750000000002
translation quality 0.00121132
training set 0.001196576
language 0.00117482
different order 0.001144019
hash functions 0.001141679
error analysis 0.0011269230000000002
significant space 0.001126134
hash map 0.001114178
frequency information 0.001102601
small number 0.001102011
feature functions 0.001091904
training scheme 0.001091294
additional memory 0.0010892879999999999
false pos 0.001086874
training procedure 0.0010831600000000001
actual error 0.001077188
other items 0.001067849
overestimation error 0.0010638380000000001
previous count 0.001061827
error rates 0.001055222
binary feature 0.0010465840000000001
minimum error 0.001045865
pos rate 0.001045503
effective error 0.0010341830000000002
test nega 0.0010318319999999999
other comparison 0.001031561
other hand 0.001028169
error guarantees 0.00102431
parallel corpus 0.0010163490000000002
other components 0.001015171
different quantisation 0.001013269
space requirements 0.001003839
gigaword corpus 0.001003612
test sen 0.001002313
possible counts 9.97875E-4
test item 9.962109999999999E-4
optimal number 9.89213E-4
membership probability 9.8514E-4
first set 9.84789E-4
corpus europarl 9.77264E-4
feature func 9.70259E-4
rate figure 9.68128E-4
mial feature 9.6022E-4
gzip size 9.590690000000001E-4
such lms 9.55483E-4
frequency bloom 9.55325E-4
bit array 9.50769E-4
event space 9.50557E-4
space savings 9.49882E-4
quantised count 9.488470000000001E-4
space overheads 9.42596E-4
discounted count 9.396270000000001E-4
different sizes 9.37856E-4
false positive 9.296969999999999E-4
