Selecting Attributes for Opinion Mining in Different Domains

P Kalaivani

Abstract


Sentiment Classification is an important tool to handle and summarize the general opinions about multi domain reviews, movie reviews and music reviews.  Various machine learning algorithms have been studied in previous literatures. In this paper, we evaluate the accuracy of movie reviews and multi domain reviews and We used Information Gain, Chi Squared and Weight by Support Vector feature selection methods and TF-IDF weighting scheme along with classification algorithms such as KNN, SVM, SVM-PSO and NB. We compared the accuracy of SVM with and without feature selection. An empirical result shows that SVM approach with SVMW feature selection method outperformed other classification algorithms.


Keywords


Sentiment Classification, Machine learning algorithm, Feature selection, SVM.

Full Text:

PDF

Refbacks

  • There are currently no refbacks.


Subscribe to Print Journals

 IJAIKD is currently Indexed By   


 http://rgjournals.com/public/site/images/mittalberi/scholar_logo_lg_2011.gif  Journal Seek