SENTIMENT ANALYSIS FOR NYKAA WEBSITE USING SVM, TWIN SVM, NAÏVE BAYES, AND LSTM

Abstract

Many websites such as Twitter, blogs, and e-commerce sites are popular nowadays, which display a tremendous amount of information about various topics, such as reviews and discussions on events. To manually try to understand the essential opinion regarding something is very time consuming. Opinion mining or sentiment analysis is used, which automatically analyzes text using machine learning approaches and tries to give the idea of people's sentiments regarding a topic or product. Nykaa is one of the leading online shopping websites, where a large amount of information is available. The paper uses sentiment analysis on the Nykaa dataset, where we train the machine to generate the ability to define the overall opinion about a particular context, such as negative or positive. The input data is first pre-processed, then this minimalized data is converted into vector space as the machine understands numbers, not text, using sentiment score. Then machine learning algorithms such as Naïve Bayes, SVM, Twin SVM, and LSTM are applied, and results are evaluated and compared where LSTM performs better.