Amazon is using artificial intelligence to combat fake product reviews and inflated star ratings.
It is employing a new AI machine-learning system that the online retailer built in-house to boost the prominence and weight of verified customer purchase reviews, those marked as helpful by other users and newer, more up-to-date critiques on its site.
The system will bring what Amazon thinks are more accurate reviews to the top and use them to create a star rating.
Previously star ratings were simply an average of all reviews, which allowed fake reviews to heavily influence the first-glance rating even if verified purchasers had slated the product.
The algorithm will improve over time, Amazon told technology site Cnet. Its first effects may not be visible for some time as the work only began on Friday.
Customer reviews have become the cornerstone of trust in the online shopping world. Where users cannot see in person what the products are like before they buy them, the ratings and reviews of users who have supposedly bought them before can make or break a product.
Can you trust that five-star review?
That means marketers have taken to attempting to influence star ratings, especially in the initial stages of a product going on sale on any particular site. They post fake, inflationary reviews or pay users to do so on their behalf.
The practice known as “astroturfing” – fake grassroots campaigns – is widespread across a variety of sites and services. Amazon, as one of the world’s largest online retailers, is a significant target.
But Amazon also indicated that its system will be capable of differentiating between original products and those modified to fix issues or customer complaints sold through the same listing.
The reviews of the fixed product will be used to outweigh older ones of the broken product to create a more consistent and useful rating of the product users would receive now.
Astroturfing is not confined to retailers. TripAdvisor and other travel sites are also heavily affected by the practice.
“We have a whole team dedicated to the problem of fake reviews,” James Kay from TripAdvisor UK told the Guardian. “We use both automated systems and a dedicated team to review reviews and weed out fake entries.”
To maintain user trust, more and more sites that host user reviews will have to implement smarter and smarter systems to prevent abuse.
Amazon’s AI-based system will be closely watched by both competitors and creators. If it works it could end astroturfing as we know it today.
Comments (…)
Sign in or create your Guardian account to join the discussion