training data 0.00427055
test data 0.003348912
corpus data 0.003143672
data sets 0.002909027
training method 0.002862176
english data 0.002852969
small data 0.0028204239999999998
training set 0.00279678
much data 0.00264792
loc data 0.002641466
suitable data 0.002626632
dutch data 0.00261632
tuna data 0.0026147049999999997
missing data 0.002612662
training sets 0.002444457
english training 0.0023883990000000003
small training 0.002355854
specific training 0.00221723
full training 0.0022085
much training 0.00218335
training instance 0.0021740880000000002
entire training 0.002161801
dutch training 0.00215175
atic training 0.002142708
different test 0.0020124419999999997
different set 0.00192488
training 0.00190299
test set 0.001875142
english test 0.001466761
different cost 0.0014154929999999999
selection algorithm 0.0013822740000000002
good performance 0.001357333
selection results 0.001290776
same language 0.001285339
other attribute 0.001284129
different pattern 0.001281761
development set 0.001250464
set size 0.001247237
overall performance 0.001237638
dutch test 0.001230112
performance ceiling 0.001220907
chosen method 0.001204921
full set 0.0011993
set sizes 0.001191404
cost function 0.001186369
reg algorithm 0.001164872
entire set 0.001152601
attribute selection 0.0011447739999999999
other reg 0.001135047
clustering algorithm 0.0010914990000000001
other attributes 0.001065525
same english 0.0010642260000000001
language costs 0.001060479
first experiment 0.001051107
incremental algorithm 0.00103302
dutch corpus 0.001024872
tuna corpus 0.001023257
transparent corpus 0.001021061
corpus frequencies 0.0010173860000000001
mean results 0.001008707
people language 0.00100252
other objects 9.934E-4
furniture domain 9.90895E-4
performance 9.77543E-4
people domain 9.67389E-4
method 9.59186E-4
accuracy scores 9.54431E-4
selection algorithms 9.45704E-4
domain complexity 9.416170000000001E-4
few properties 9.339209999999999E-4
encouraging results 9.27792E-4
consistent results 9.26351E-4
domain objects 9.18889E-4
cost functions 9.177090000000001E-4
given domain 9.11668E-4
evaluation experiments 8.85767E-4
selection perfor 8.72416E-4
same frequency 8.53705E-4
complete ranking 8.53426E-4
same type 8.39691E-4
such corpora 8.24706E-4
evaluation metric 8.22223E-4
ing instances 8.193079999999999E-4
frequency information 8.18502E-4
good news 8.134590000000001E-4
time con 8.11613E-4
next experiment 8.070449999999999E-4
current evaluation 8.0689E-4
ferent sets 8.048090000000001E-4
function 8.01966E-4
mapping properties 7.982479999999999E-4
much time 7.89949E-4
acceptable attribute 7.873209999999999E-4
english descriptions 7.84685E-4
tribute sets 7.81392E-4
bad sets 7.81392E-4
racy scores 7.79223E-4
frequent properties 7.78356E-4
our english 7.76344E-4
corpus 7.76112E-4
