Amazon using machine learning for product reviews

Amazon is the largest online retailer in the world, serving almost 244 million active users. With such a large user base reputation is key which is why Amazon prides itself on its product reviews. These reviews are about to get a major facelift, but first, lets talk about the challenges.

When looking for a product online its hard to tell whether we are spending our money wisely or not. The product reviews from previous users are there but they are just too large in number. You don’t know which one to trust and which one to ignore. Also, the product rating is the average of ALL the product reviews, even the inaccurate ones. These problems is what led Amazon to try a Machine Learning software in the USA.

This software allows assigns preference to different types of reviews. Newer reviews get preference over older ones, reviews from verified Amazon customers get more preference than reviews from no-verified customers and reviews that get ‘up’ votes (for being helpful) get more preference than reviews that don’t get votes. Also, the product rating is now the ‘weighted’ average of the reviews where more weight is given to the ‘preferred’ reviews (i.e. the ones stated above).

Amazon hopes that this new system will give users access to more relevant content and enhance their overall experience on the site.

Ishtiaqur Rahman

Contact

Posts

163