MENU

Amazon Taps Machine-Learning System To Keep Customer Reviews In Check: Here Are The Changes

Close

Amazon has made some enhancements to the algorithm of its review system through a machine-learning platform to display the most recent and most reliable Amazon reviews and also updated ratings on product pages.

With these algorithm tweaks, Amazon will now put more emphasis on the verified and freshest customer reviews written through comments to determine the top reviews to show.

The change also aims to improve the calculation of the product's overall rating through the website's 5-star rating system to reflect the present product experience. An initially high-rated appliance's grade should change if verified purchasers will post new negative comments about the quality of the appliance after weeks of usage. Amazon previously displayed a product's score based on pure averages of all reviews.

Amazon incorporated the changes last Friday in the U.S. and should be visible in two primary ways. First, the machine-learning platform will rank and continuously adjust the order of the reviews being displayed on the website according to the ongoing assessment of which customer reviews are most important. Secondly, the overall star rating of the product will now take into consideration the helpful votes of the customers, whether the reviewers are verified buyers, and the age of the written review.

Amazon is also testing a system that will allow customers to score a product with a numerical rating based on its specific attributes such as durability and battery life.

Since customers cannot test or touch Amazon products before purchasing, most Amazon buyers often rely on the displayed 5-star rating system and read the written reviews of the other customers for the quality of the item. For over 20 years, customer reviews have been a significant contributor to Amazon's success in online product selling.

Amazon spokesperson Julie Law stated that the new platform was monitored closely before its implementation and the system will improve over time based on its capability to learn continuously from all the customer reviews encoded. This latest move also aimed to crack down on fake reviews on the product pages.

Last April, Amazon filed a lawsuit against a group of site operators that posted fake 4-star and 5-star customer reviews of their products so that Amazon customers would purchase them.

The exact grading system of the machine-learning platform in computing product scores, based on the new factors considered, would most likely remain a trade secret to avoid vendors from cheating the system.

ⓒ 2018 TECHTIMES.com All rights reserved. Do not reproduce without permission.
Real Time Analytics