Product Review Assistance Based on Feature Sentiment Analysis Using NLP

Abstract

In today's digital era, customers face challenges in making confident purchasing decisions due to the overwhelming volume of online product reviews. Simultaneously, manufacturers struggle to extract meaningful insights from this vast amount of feedback. There is a pressing need for a system that can efficiently analyze product reviews, offer valuable insights, and bridge the gap between customers and manufacturers, enabling informed choices and product enhancements in the real world. As a solution to this problem, in this work we propose a method which analyses the product reviews along with the extraction of product features and in result defining the overall sentiment towards the product by the consumers with respect to its good and bad features. The proposed methodology includes use of custom dataset with the help of web scrapping (flask), dependency parsing, sentiment prediction of reviews using BERT and matrix method to determine the feature sentiment.