Sentiment Analysis for Women's E-commerce Reviews using Machine Learning Algorithms

Abstract

The opinion of others is significant in the decision-making process of a person. It is common for websites to have a review section under their products where people express their opinions, enabling prospective consumers to evaluate the items. Hence, analysis of these reviews plays a major role in e-commerce. The branch of artificial intelligence dealing with the analysis and classification of positive and negative opinions is Sentiment Analysis or Opinion mining. This paper details how sentiment analysis has been performed on women's e-commerce reviews from Aamazon.com using Weka. The dataset was pre-processed, and various other functionalities were performed for better classification. The most suitable classifier for sentiment analysis from the following: Naïve Bayes, JRip, J48, and Sequential Minimal Optimization (SMO), from four different categories (Bayes theorem, Rules, Trees, Support Vector Machines) used in combination with boosting algorithm AdaBoost, will be determined by comparing and analyzing their results.