Aspect-level sentiment analysis on goods and services tax tweets with dropout DNN
by E. Deepak Chowdary; S. Venkatramaphanikumar; K. Venkata Krishna Kishore
International Journal of Business Information Systems (IJBIS), Vol. 35, No. 2, 2020

Abstract: Sentiment analysis (SA) is a primary use case for natural language processing, where data scientists analyse comments on social media to get instant feedback and improvement in future product releases. In this study, SA was performed on the goods and services tax (GST), which is one of the greatest tax reforms in India. In this study, crucial statements (tweets) about GST were analysed using POS-N-gram tokenisation approach to extract word tokens for classifying sentiments or opinions. The objective of this proposed work is to improve the sentiment classifications accuracy of the review data with an optimal number of reduced terms. In this study, a novel approach improved PCA is proposed for dimensionality reduction and a dropout deep neural network classifier is proposed for sentiment classification. These methods are evaluated on the proposed dataset (Corpus-1) and two other benchmark datasets like movie reviews (Corpus-2) and SemEval 2016 (Corpus-3) datasets. Experimental results clearly evident that the proposed approach outperforms the existing methods.

Online publication date: Thu, 08-Oct-2020

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Business Information Systems (IJBIS):
Login with your Inderscience username and password:

Username:        Password:          Forgotten your password?

Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com