Polarity Prediction on Product Reviews Based on Classification Algorithms

Abstract

Sentiment analysis is analyzing customer’s opinion whether they are positive or negative on the product they purchased. It is opinion mining which is major task of Natural Language Processing. The main objective of sentiment analysis is to categorize the data based on product review data’s polarity, it majorly consists of three steps document level, sentence level, aspect level. Sentiment analysis is mostly grounded on supervised learning in machine learning. Online product reviews are collected and analyzed. Online platform of shopping is very much helpful to gather review or opinion of customers to analyze. A sentiment sentence is categorized as positive or negative review based on the type of words used in the sentence, where the sentence is tokenized into separate English words to get analyzed. Various demonstrations were performed on both sentence-level and review-level categorizations with the desired output.