Comparison of Sentiment Analysis using VADER and RoBERTa

Abstract

Analysing human emotions about a product or an incident is called sentiment analysis. In the current world where there are uncountable number of platforms where people free express their feelings, there is an overload of sentiment based data available. It has become a challenge to gather this data and get some useful information out of it. Sentiment analysis has become an important domain even before the time when various sophisticated techniques of machine learning and deep learning were available. By analyzing social media, product reviews, and customer support interactions, companies can promptly respond to emerging trends, manage their brand reputation, and enhance customer satisfaction. This paper makes an effort to analyze the sentiments of people about a product in Amazon. This paper first uses very basic machine model called VADER (Valence Aware Dictionary and sEntiment Reasoner) and then a more sophisticated Deep Learning approach called RoBERTa (Robustly optimized Bidirectional Encoder Representation from Transformers approach).