Enhanced Sentiment Analysis Technique using Machine Learning (B.R.A.G.E technique)

Abstract

Sentiment Analysis have been the most growing topic in the recent years. It is the use of text analysis to examine the opinion or attitude towards a topic. In the past years, there have been a significant growth in the volume of research on Sentiment Analysis, on different detection level such as document level, sentence level and feature level. One of the famous existing sentiment analysis models is Naïve Bayes, a supervised machine learning model. In this study, we identified that the existing Naïve Bayes model trained and tested with incident/accident-related dataset gave an accuracy level of 71%. Additionally, this study describes how the proposed B.R.A.GE. technique has slightly enhanced the sentiment analysis prediction accuracy using incident/accident-related dataset. In conclusion, the proposed B.R.A.G.E technique has not significantly improved the accuracy but hence could be further improvised.