sentiment classification 0.003300959
negative sentiment 0.003159865
different sentiment 0.003107073
positive sentiment 0.00303285
ing sentiment 0.0030290250000000003
sentiment analysis 0.0030005730000000003
level sentiment 0.002948467
sentiment label 0.00283029
sentiment resources 0.002830259
sentiment value 0.0028223280000000002
sentiment decision 0.0027778860000000002
sentiment categorization 0.002766555
tive sentiment 0.0027656860000000003
overall sentiment 0.002760638
sentiment infor 0.002744793
prevailing sentiment 0.002708349
sentiment 0.00248714
review data 0.0017685910000000001
polarity classification 0.001704646
feature set 0.001483054
negative words 0.001449625
other words 0.001442169
movie review 0.001433671
classification problem 0.0014134080000000001
human rationales 0.0013219529999999998
negative rationales 0.001307322
polarity values 0.001301714
polarity lexicon 0.001300973
training corpus 0.001300972
good rationales 0.0012898340000000001
svm learning 0.001288942
same feature 0.001257535
view data 0.001254064
training example 0.001241004
training examples 0.001225524
different domain 0.0012246840000000002
review dataset 0.001211295
human annotation 0.001204503
opinion words 0.00120297
such sentence 0.001194475
unigram features 0.001163101
categorization performance 0.001156109
contextual polarity 0.0011556169999999998
previous research 0.001151159
balanced performance 0.001138605
polarity lexicons 0.001135495
positive document 0.001133449
performance gains 0.00112783
contrastive training 0.00112076
ditional training 0.001113702
same movie 0.001106009
classification tasks 0.001083076
original review 0.001082611
different parameter 0.001081062
first approach 0.001080085
learning models 0.001075355
other approaches 0.001073276
product review 0.00107307
negative reviews 0.001072952
movie reviews 0.001062737
human annotator 0.001061422
classification accu 0.001047571
ing approach 0.0010463159999999998
classification boundary 0.001034677
learning methods 0.0010315440000000001
same document 0.0010312379999999999
standard svm 0.001021688
single word 0.001018261
annotator rationales 0.001008663
uct review 0.00100057
learning method 9.98479E-4
research community 9.972079999999999E-4
learning experiments 9.87331E-4
test set 9.845E-4
random rationales 9.82103E-4
human annotators 9.754519999999999E-4
negative ratio 9.7125E-4
negative sen 9.666329999999999E-4
annotated rationales 9.65783E-4
svm light 9.58361E-4
contextual information 9.56663E-4
supervised learning 9.53363E-4
svm constraints 9.51381E-4
good rationale 9.47765E-4
subjective sentences 9.453630000000001E-4
few rationales 9.44429E-4
negative terms 9.37607E-4
such rationale 9.33425E-4
human annota 9.275759999999999E-4
noisy rationales 9.25416E-4
automatic rationales 9.24564E-4
text span 9.19604E-4
human effort 9.187259999999999E-4
learning algorithm 9.16562E-4
human anno 9.10431E-4
opinionfinder rationales 9.09771E-4
human intu 9.09081E-4
text categorization 9.05783E-4
sophisticated rationales 9.02368E-4
negative opin 8.94513E-4
