The topic This AI can tell a real online review from a fake one, and it’s surprisingly accurate is currently the subject of lively discussion — readers and analysts are keeping a close eye on developments.
This is taking place in a dynamic environment: companies’ decisions and competitors’ reactions can quickly change the picture.
Fake reviews are a real menace for online shoppers. If you have ever bought something online based on glowing reviews only to receive a disappointingly subpar product, you know what I mean. A new study published in the International Journal of Information and Communication technologies proposes an AI-powered system that can not only detect fake reviews, but also trace how they spread.
Most existing fake review detection systems focus on the text of a review. That approach worked for a while, but fake reviewers have gotten smarter. They now pair carefully written text with misleading images to make their reviews look authentic. Text-only tools struggle to catch this, and that’s a real problem for shoppers and honest sellers alike.
The researchers addressed this by building a system that looks at multiple signals at once. It analyzes the review text using two different methods, a text convolutional neural network and pre-trained language models, to capture both surface-level and deeper meaning in the words. It also factors in reviewer behavior, since fake accounts tend to have default profile pictures and system-generated usernames, unlike real users who tend to personalize their accounts.
The short answer is yes. Review images are analyzed separately using a residual network, a type of deep learning tool commonly used for processing visuals. Once all these signals are gathered, the system fuses them together to make a final call on whether a review is genuine.
When a review is flagged as fake, a Transformer model kicks in to map its origin and track how far it spread through the network.
Tests on a large dataset from JD.com showed that the system achieved a recognition accuracy of 94.2% and a tracing accuracy of 93.5%, outperforming all existing methods it was compared against. This kind of accuracy could eventually mean fewer misleading reviews and more trustworthy ratings to shop by.
