"Megamarket" has learned to filter reviews from neural networks and bots

2024: Implementation of feedback detection model

The megamarket has learned to filter feedback from neural networks and bots. Megamarket announced this on July 16, 2024.

The implemented model will practically reduce to zero reviews written by bots and neural networks. Now Megamarket users can be sure they are reading reviews that stream the real-life experience of shoppers. And sellers, when calculating increased bonuses for reviews, that they will be written consciously, without the use of neural networks.

This technology allows you to filter reviews (both positive and negative) without specific details that can be placed under any product. At the same time, the model is trained so that the probability of accepting a comment from a real user for AI text is minimal. Users can be in no doubt that they are reading reviews from buyers: the algorithm does not miss AI comments and removes those that were written earlier. Sellers will be able to assign increased bonuses for reviews and be sure that they will be written consciously, without the use of neural networks. This approach will allow you to quickly fill product cards with high-quality reviews, increase conversion and get high-quality feedback, which will help refine the product and other processes.

Шаблон:Quote 'author = said Daniil Boldin, product manager of Megamarket product card.

As a result of the introduction of this technology, reviews left from 2020 to July 2024, generated using AI technologies, were identified and removed. The marketplace plans to iteratively improve the quality of the model.

Шаблон:Quote 'author = noted Artem Gerasimovich, head of content marketing at Megamarket.