same word 0.001649624
same data 0.0015088979999999998
particular word 0.0013001599999999999
features 0.00120503
training set 0.00107139
voting function 0.001069176
other classifiers 0.001063372
different classification 0.001047894
weighting function 0.001027177
neighbor algorithm 9.887239999999999E-4
other presidents 9.46108E-4
such corpora 9.36892E-4
such filters 9.23823E-4
same attribute 8.97616E-4
classification errors 8.836849999999999E-4
same direction 8.810339999999999E-4
different approaches 8.42478E-4
accuracy estimation 8.28285E-4
further work 8.26214E-4
same members 8.24723E-4
function 8.14937E-4
different part 8.12211E-4
bayes classifier 8.08366E-4
spam corpus 8.065430000000001E-4
public corpus 7.88458E-4
initial results 7.88157E-4
spam messages 7.84149E-4
different attributes 7.84023E-4
different filter 7.79662E-4
preliminary results 7.73295E-4
overall performance 7.68166E-4
learning algorithms 7.644030000000001E-4
optimal results 7.630499999999999E-4
weighted accuracy 7.52274E-4
different president 7.52161E-4
algorithm 7.4645E-4
message class 7.454079999999999E-4
generalization approach 7.43208E-4
machine learning 7.38563E-4
classifier ensemble 7.36261E-4
classifier ensembles 7.33592E-4
additional errors 7.335390000000001E-4
results tables 7.32413E-4
experimental results 7.319259999999999E-4
spam message 7.288959999999999E-4
different misclassification 7.2616E-4
corresponding performance 7.24255E-4
evaluation measures 7.1484E-4
first version 7.10656E-4
indicative results 7.08841E-4
common errors 7.055410000000001E-4
benchmark corpus 6.99595E-4
small values 6.99291E-4
holdout accuracy 6.98962E-4
incoming messages 6.89098E-4
learning techniques 6.80277E-4
multiple classifiers 6.76655E-4
empirical evaluation 6.75349E-4
new instances 6.71905E-4
unsolicited messages 6.668080000000001E-4
support vector 6.65488E-4
legitimate messages 6.62198E-4
vector machines 6.58596E-4
further improvement 6.5609E-4
linguist messages 6.50217E-4
several configurations 6.44876E-4
message category 6.44815E-4
average werr 6.42173E-4
model 6.40358E-4
text categorization 6.39417E-4
extra work 6.359930000000001E-4
public approximation 6.34997E-4
information gain 6.34307E-4
incoming message 6.338450000000001E-4
much space 6.2941E-4
error types 6.28699E-4
vector nxxxxx 6.22938E-4
average percentages 6.2265E-4
legitimate message 6.06945E-4
weighted error 6.01272E-4
tcr values 6.00611E-4
true class 5.98131E-4
spam precision 5.98112E-4
level classifiers 5.90301E-4
ground classifiers 5.87477E-4
incoming spam 5.86541E-4
cost scenario 5.85466E-4
previous experiments 5.822469999999999E-4
error rate 5.812E-4
cost ratio 5.77113E-4
previous research 5.70677E-4
low cost 5.703150000000001E-4
individual decisions 5.66792E-4
training 5.66412E-4
decision trees 5.65906E-4
misclassification errors 5.61951E-4
strong distance 5.61428E-4
base form 5.57028E-4
public benchmark 5.56559E-4
total cost 5.55464E-4
