Achievements

Research Article: Analysing user sentiment of Indian movie reviews: A probabilistic committee selection model. 29-Mar-2018

Prof. Shrawan Kumar Trivedi

Assistant Professor (IT & Systems)

Article: Analysing user sentiment of Indian movie reviews: A probabilistic committee selection model.

Journal: The Electronic Library. [B class as per ABDC list, Scopus Indexed, Impact factor - 0.484]

Abstract:

Purpose: To be sustainable and competitive in the current business environment, it is useful to  understand  users’  sentiment  towards  products  and  servicesThis  critical  task  can  be achieved  via  natural  language  processing  and  machine  learning  classifiers.  This  research proposes a novel probabilistic committee selection classifier (PCC) to analyze and classify  the sentiment polarities of movie reviews.
Design/methodology/approach: An 
Indian  movie  review  corpus  is  constructed  for  use  in this research. Another publicly available movie review polarity corpus is also involved with regard  to  validating  the  resultsThe  greedy  stepwise  search  method  is  used  to  extract  the features/words of the reviews. The performance of the proposed classifier is measured using different  metricssuch  as  F-measure,  false  positive  ratereceiver  operating  characteristic (ROC)  curve,  and  training  time.  Further,  the  proposed  classifier  is  compared  with  other popular  machine-learning  classifierssuch  as  Bayesian,  Naïve  Bayes,  Decision  Tree  (J48), Support Vector Machine, and Random Forest.. 
Findings: The results of this study show that the proposed classifier is good at predicting the positive or negative polarity of the movie reviews. Performance accuracy and the value of the ROC curve of the PCC is found to be most suitable
of all other classifiers tested in this study. This  classifier  is  also  found  to  be  efficient  at  identifying  positive  sentiments  of  reviews, where it gives low false positive rates for both Indian Movie Review and Review Polarity corpora used in this study. The training time of the proposed classifier is found to be slightly high from Bayesian and Naïve Bayes and J48. 


Keywords: Probabilistic  committee  selectionSentiment  analysisIndian  movie  review, Machine learning classifiers, Greedy stepwise search method.