Automatic Product Review Sentiment Analysis Using Vader and Feature Visulaization

Abstract

Sentiment polarity analysis has been a popular research field for data, scientists over the last decade. Movie reviews, hotel reviews, social media like twitter reviews and product reviews have been the subjects of sentiment polarity analysis. NLTK has been facilitating these researchers with necessary classification tools to verify and finetune the accuracy of sentiment polarity analysis models. The most interesting part of the research is the sentiment polarity using the intensity of the sentiments in the reviews. The Vader sentiment analysis tool is one such tool which uses a specially developed lexicon to classify the sentiment based on the intensity of sentiments. Vader also facilitates unsupervised sentiment analysis, unlike other supervised machine learning techniques. This study explores Vader tool for unsupervised and online sentiment analysis of product reviews. The study also focusses on the domain based training datasets and their universal applicability for sentiment classification. Finally, the study highlights the usefulness of direct visualization techniques for selected high frequency negative and positive feature sentiments