eduzhai > Applied Sciences > Engineering >

Improved Multinomial Naïve Bayes Approach for Sentiment Analysis on Social Media

  • Save

... pages left unread,continue reading

Document pages: 7 pages

Abstract: Due to a substantial growth in amount of internet users over the years, there has been a large growth in amount social media users. Every person likes to voice their opinion on some matter that concerns them directly or indirectly. With the ease of expressing our views to millions of people in just a matter of few seconds by a few strokes of keyboard, social media has become a platform that can be efficiently utilized to extract valuable information about people. Comments made by users can be extracted using some API and then be fed to an algorithm that determines whether it is positive or not, which is the sentiment of the user. One such proposed algorithm is Multinomial Naïve Bayes Classifier which after certain steps of linguistic preprocessing gives an appreciable result when compared to more sophisticated algorithms on previously unseen data even on a very small training set containing less than 1000 training examples. We observe how generative models like Naïve Bayes usually perform better than discriminative models like SVM and Decision Trees for the task of Sentiment Analysis on micro blogging websites where the corpus is small.

Please select stars to rate!

         

0 comments Sign in to leave a comment.

    Data loading, please wait...
×