Survey of Product Reviews using Sentiment Analysis

Background/Objective: Online shopping encompasses large variety of products and reviews which gives rich and valuable source of information for both enterprise and user. The review available on the internet is often disorganized that makes the user a difficulty in navigating the information and gaining knowledge. The aim of this study is to predict the users' opinion of a product based on their online reviews. Methods: We propose a technique called semantic orientation, which automatically finds the frequently used terms for an aspect of a product from online customer reviews and other important context considered here is a dynamic dataset. Firstly the product aspect is identified, and then sentiment classification is done. Additionally, other techniques like stop words removal, context based mining and stemming is employed. Findings: It provides an efficient way of predicting the user's opinion and thereby suggesting them. Applications/Improvements: The proposed system has been tested in various products and is analysed. This system can be implemented in real world application like predicting opinions of the user. Hence improving the user's accessibility.