word sense 0.00323234
sense class 0.002487146
other features 0.002402427
training data 0.0023260629999999997
data set 0.002300153
large sense 0.002175718
sense disambiguation 0.00215469
test data 0.002127034
frequent sense 0.002118426
sense assignment 0.002031385
sense disambigua 0.002021582
example features 0.002011576
same data 0.0019857399999999997
feature value 0.001961164
collocation features 0.0019438510000000001
only features 0.0018692070000000001
feature values 0.001843683
common data 0.001821196
learning algorithm 0.001815338
training set 0.0017853959999999999
word wall 0.001766466
sense 0.00176166
possible feature 0.001750523
word brown 0.0017369
word occurrences 0.001730269
particular data 0.001700759
feature weighting 0.001680918
data consortium 0.001672481
ambiguous words 0.001638348
feature weights 0.0016367460000000001
only feature 0.0016257770000000001
different learning 0.001609907
training corpus 0.001587664
test set 0.001586367
feature representation 0.001578849
symbolic feature 0.00154133
features 0.00152771
training examples 0.0014908859999999999
training example 0.001389519
other machine 0.001353621
words 0.00135296
machine learning 0.001338192
learning approach 0.001309903
test examples 0.001291857
feature 0.00128428
neighbor algorithm 0.00127956
algorithm pebls 0.001233605
matching training 0.001218924
training exam 0.001203064
additional set 0.001193639
test example 0.00119049
learning algorithms 0.00119002
particular training 0.001186002
sentence form 0.0011847749999999999
learning problems 0.001170255
learning approaches 0.001153995
other assignment 0.001144442
other hand 0.001144351
rule set 0.001134691
informative set 0.001133566
holdout set 0.001132157
second test 0.001130689
other variations 0.001129697
disambiguation accuracy 0.001125195
chine learning 0.0011117570000000001
classification performance 0.001098507
large corpus 0.001096069
ing examples 0.001095195
large value 0.0010909420000000001
first test 0.001079198
morphological form 0.001068025
pruning method 0.001055776
assignment method 0.001051035
optimal parameter 0.0010486
majority class 0.001028494
accuracy figures 0.001025291
test sets 0.001021844
maximum value 0.001015783
average number 0.001003488
cal form 0.00100007
isambignation accuracy 9.99691E-4
phological form 9.92487E-4
disambignation accuracy 9.89912E-4
ambiguation accuracy 9.89408E-4
accuracy fig 9.874179999999999E-4
biguation accuracy 9.85194E-4
frequent classifier 9.79847E-4
collocation values 9.755440000000001E-4
experimental results 9.736300000000001E-4
default value 9.73541E-4
wsj corpus 9.69476E-4
value difference 9.609379999999999E-4
algorithm 9.5605E-4
value pruning 9.5135E-4
brown corpus 9.482310000000001E-4
combined corpus 9.47788E-4
local collocations 9.463570000000001E-4
parameter setting 9.43982E-4
parameter settings 9.34014E-4
large number 9.319E-4
