D10-1037 similarly uses a bootstrapped local polarity classifier to identify sentence polarity
D13-1187 F-measure ranging from 0.5 - 5 % . The polarity classifiers overall achieve much higher scores
D15-1016 in steps of 0.01 . + ve / - ve polarity classifier , the accuracy showed an improvement
J11-2001 methods to build domainindependent polarity classifiers . Read and Carroll built different
N12-1085 techniques such as Nakagawa et al. 's polarity classifier , they function as a baseline
D15-1019 experimental results for the simile polarity classifier , using both manually annotated
P04-1035 Naive Bayes . Figure 4 shows the polarity classifier results as N ranges between 1
P04-1035 the polarity dataset . Default polarity classifiers We tested support vector machines
D09-1062 . We suspect it is because the polarity classifiers we experimented with is not highly
D09-1062 two competitive expression-level polarity classifiers from 64.2 % - 70.4 % to 67.0
J09-3003 the results for the all-feature polarity classifiers are also given . Inter - estingly
J09-3003 together ? For all algorithms , the polarity classifier using all the features significantly
D09-1061 order to train a high-performance polarity classifier . Some recent attempts have been
D15-1016 obtained for both ratings based and polarity classifier on RHMR and MRD Dataset . Our
P04-1035 above , we can use our default polarity classifiers as " basic " sentencelevel subjectivity
D11-1015 NTCIR-8 Chinese MOAT as the baseline polarity classifier ( BPC ) in this paper . Error
P04-1035 we give as input to the default polarity classifier an extract consisting of the
P04-1035 ) . Each default documentlevel polarity classifier is trained and tested on the
D09-1020 subjective F-measure . 4.2.3 Contextual Polarity Classifier We now apply SWSD to contextual
D15-1016 each setting for both Ratings and Polarity classifiers . We also report 5-fold cross
hide detail