Large Korean online retailers such as 11Street and eBay Korea have employed artificial intelligence to comb out fakes or paid consumer reviews amid increasing power of SNS influencers over consumer behavior.

According to sources on Tuesday, 11Street, one of the country’s top online marketplaces, internally developed an AI robot engine to automatically filter out reviews full of repeated words such as `Delicious, Delicious, Delicious’ and other unfaithful texts. The company’s AI system can also monitor video reviews round the clock.

Unlike the U.S. retailer system, it is difficult for consumers to use a fake ID on 11Street because users can leave a ‘Like’ once per ID, and have to verify themselves to post a review after purchasing products, said Kim Yong-hee, head of the company’s social media planning team.

eBay Korea, the country`s largest e-commerce company, with online shopping malls such as Gmarket, Auction and G9 under its arm, only real purchasers to leave posts on the review corner.

Amazon attaches its “Amazon’s Choice” badge to certain products based on the number of consumer reviews and listings, but the system is often criticized as it can be manipulated by sellers to receive the endorsement.

Although Gmarket and Auction provide a product sorting function with `highest ratings`, they have no system that puts tags such as “Amazon’s Choice” to popular products or concentrates exposure on highly rated products.

eBay Korea also uses an AI system to filter out reviews that consist of unrelated images or texts with vague content. A fraud detection system is also in place to monitor fake reviews in a variety of ways, such as whether sellers and buyers are matched or relevant, and there are no abnormal patterns. For example, the system compares and analyzes the seller’s sign up information and access information and the buyer’s counterpart data to calculate their possible connection.

By Kim Gi-jung and Minu Kim

[ⓒ Pulse by Maeil Business Newspaper & mk.co.kr, All rights reserved]