In this digital age, anything and everything could be accomplished virtually. There are countless businesses that could help one finish certain tasks they want. The convenience and swiftness are specifically true when it comes to online shopping, and consequently product reviews as well. The extranet makes it so that there could be up to thousands of reviews for each product, each with its own context, rendering it nearly impossible to manually assess each one individually. One could count on the scores of the products for analysis, but it often only provides a surface-level overview. In order to address this gap, we set out to draw out the relationship between product reviews and score by analyzing the sentiment of the reviews by applying sentiment analysis.
Specifically, our goal is to build deep learning models that can learn from the text within the customer reviews and accurately predict the product’s score. The process includes data cleaning, tokenization, vectorization, and training deep learning models, models testing and evaluation. This approach provides a deep understanding of product reviews beyond numerical scores.