Aspect Based Sentiment Analysis on Product Reviews

Abstract

With the popularity and growing availability of opinion rich sources such as reviews from e-commerce sites, choosing the right product from huge product brands have difficult for the user. In order to enhance the sales and customer satisfaction, most of the sites provide opportunity for the user to write review aspects about the product. These reviews are in text format and increases day by day. It is difficult for the user and manufacturer to understand likes and dislikes of a customer about the product. In this situation sentiment analysis helps the people to analyze the reviews and come to conclusion whether it is good or bad. Sentiment Analysis which also known as opinion mining is one of the subsection in Natural Language processing in which it learns about Sentiment or subjectivity from reviews. The main purpose of the project is to develop a system to extract the reviews from e-commerce site, extract aspect from the reviews and categorize reviews into positive and negative. We have implemented Sentiment analysis with unsupervised machine learning technique like uni-gram Lexicon, bi-gram Lexicon and Supervised technique like Support vector machine (SVM). These techniques are experimented on real time e-commerce dataset. Out of three techniques, SVM outperformed with an accuracy of 84%.