training data 0.002871341
unlabeled data 0.002742057
other data 0.002738683
labeled data 0.002722072
negative data 0.0027154709999999997
data points 0.002675755
ambiguous data 0.002514165
sample data 0.002486015
sentiment classification 0.00248176
annotated data 0.0024359019999999998
unambiguous data 0.002425868
data point 0.002411701
data collection 0.002357824
notated data 0.002347557
polarity classification 0.002124598
svm classifier 0.0019740870000000002
text classification 0.001914262
learning points 0.001808745
learning step 0.001789371
subjectivity classifier 0.0017713870000000001
feature space 0.001766587
classification process 0.001751682
polarity information 0.001746879
discriminative classifier 0.001723282
active learning 0.001712916
classification tasks 0.001700659
classification system 0.00169821
sentiment analysis 0.001691035
performance polarity 0.0016761179999999999
feature vector 0.001672261
transductive learning 0.001657457
larity classifier 0.001657249
sophisticated classifier 0.001644261
sulting classifier 0.00164164
unsupervised learning 0.001640491
unsupervised sentiment 0.0016363909999999998
classification datasets 0.0016342140000000002
classification errors 0.001626941
ment classification 0.001612881
classification systems 0.001609893
larity classification 0.001593019
duce classification 0.001592035
ity classification 0.001577102
ambiguous feature 0.001574505
learning spectral 0.001563532
spectral learning 0.001563532
sentiment orientation 0.0015605999999999999
learning parameters 0.001549992
learning procedure 0.001536873
sentiment dimension 0.0015067219999999998
passive learning 0.001496449
sentiment classifica 0.001488239
ensemble learning 0.001484145
approach results 0.001480016
negative reviews 0.001454715
such information 0.001440402
available features 0.0014340890000000002
feature vectors 0.001430818
irrelevant features 0.0014302920000000001
different subset 0.001410179
feature relevance 0.001408401
many reviews 0.001399989
polarity classifiers 0.0013824129999999999
sophisticated features 0.001380521
clustering approach 0.0013781940000000001
appealing features 0.001378036
different authors 0.001365106
different stages 0.001360448
classifier 0.0013545
much information 0.001337756
other text 0.001300075
unlabeled points 0.0012926119999999998
classification 0.00129027
unsupervised polarity 0.001279229
clustering algorithm 0.001277685
labeled points 0.0012726270000000001
other words 0.0012718059999999999
negative points 0.001266026
information retrieval 0.001256783
ambiguous reviews 0.001253409
relevant information 0.0012486
first step 0.001237072
movie domain 0.0012346470000000002
useful information 0.001220818
supervised polarity 0.0012127029999999999
word usage 0.001209965
unsupervised algorithm 0.00119996
contradictory information 0.001199502
learning 0.00119559
sentiment 0.00119149
common approach 0.001183987
movie review 0.001179996
step movie 0.001177983
clustering task 0.001176403
performance rises 0.001173617
svm classifiers 0.001167672
unambiguous reviews 0.001165112
ing algorithm 0.0011640280000000001
polarity classi 0.001163676
large corpus 0.001161003
