CSE498, Collaborative Design, Fall 2020
Computer Science and Engineering
Michigan State University

Founded in 1994 as an online bookstore, Amazon is the largest online retailer in the world. Amazon has seen tremendous growth and success, making history by becoming the second U.S company to be valued at $1 trillion. A key factor in Amazon’s rise to the top is their e-commerce platform, which accounted for nearly 50% of all online retail purchases last year.

At Amazon’s scale, individual products often have numerous different sellers. Each seller provides a description and specification list for their product, making it Amazon’s job to compile every seller’s contribution into a comprehensive overall product description. Issues arise when multiple sellers provide separate, sometimes conflicting, descriptions for the same product, leading to inaccurate product descriptions.

Our Maestro system combats this problem by comparing and correcting the product descriptions from many sources, including the information from the seller, the manufacturer’s website, as well as third-party websites such as Target or Walmart. Maestro collects all of this information in the background, and does not require any work from the seller.

Maestro analyzes all of this data using natural language processing (NLP) and determines the best description for all products sold on Amazon. If any difference is detected between a seller’s description and Maestro’s description, the seller is notified and given the chance to change their description.

Our Maestro application reduces the number of inaccurate product descriptions presented to customers, who get exactly what they expect. This leads to greater sales and customer satisfaction.

Maestro is built with React for the front end, AWS Lambda for the back end, pre-trained SpaCy models for NLP, Amazon S3 buckets to hold files for inaccurate product descriptions, and AWS DynamoDB for holding product data provided by Amazon.