Sentiment Ranking for Opinion Extraction by Weighted Feature Scheme

  • Sujith Prabhu
  • Mukundaram Prem Kumar
  • Sridhar Amirneni

Abstract

Abstract— Sentiment Analysis is the analysis of emotions and opinions in a given text using Natural Language Processing. It helps in understanding the user attitude and feelings by analyzing reviews/comments posted by people through social media, blogs etc. Internet availability has given us an easier way of accessing the information and expressing our opinions on various topics or products. These reviews are of immense importance to the targeted organization since they largely affect the outlook of the organization. People tend to consider these reviews when purchasing or using a product.  This paper aims to analyze the user's sentiment in the form of free text by extracting the opinionated or sentiment words from the sentences using various techniques.  They are further classified into positive or negative sentiment words and are grouped under various categories using a sentiment dictionary based on which every word will be assigned a weight. The major refinement in the proposed system is the estimation of different feature weights for evaluation of overall sentiment score and this increases the accuracy by a huge margin. The average of all the values is computed and the review is rated on a scale of 1 to 5 indicating the sentiment of the user.  The sentiment score in addition to providing the user feedback also gives us an insight into the various features and helps an enterprise to focus on improving the below average features to boost the product on the whole.

Downloads

Download data is not yet available.
Published
Mar 30, 2017
How to Cite
PRABHU, Sujith; KUMAR, Mukundaram Prem; AMIRNENI, Sridhar. Sentiment Ranking for Opinion Extraction by Weighted Feature Scheme. International Journal of System Modeling and Simulation (ISSN Online: 2518-0959), [S.l.], v. 2, n. 1, p. 7-13, mar. 2017. ISSN 2518-0959. Available at: <http://www.researchplusjournals.com/index.php/IJSMS/article/view/262>. Date accessed: 22 june 2017. doi: http://dx.doi.org/10.24178/ijsms.2017.2.1.07.