VK introduced a new algorithm that considers future preferences

AI researchers at VK Research have developed a new approach that expands the capabilities of classical recommendation systems. According to IXBT.com, this algorithm can evaluate not only the user's current interests, but also how recommendations will affect their future preferences. This is reported by Ixbt.com reports.

Although neural network models currently play a major role in the operation of most digital services, classical recommendation algorithms remain relevant in areas where data processing speed and saving computational resources are important. Designed specifically for such systems, the new method serves to improve the overall quality of the content selection chain.

Technical advantages of the new approach

The researchers note that the developed solution can be successfully integrated into existing recommendation systems without fundamentally changing their architecture. This allows companies to significantly improve the quality of recommendations without rebuilding their software.

The new approach has already begun to be applied in solving a number of practical tasks within the VK ecosystem. This has demonstrated the technology's effectiveness not only in theory, but also in a real industrial environment.

Test results and international recognition

During tests conducted by specialists, this approach was evaluated on several open datasets such as VK-LSVD. The test results showed that the new algorithm allows for much more accurate and precise content selection compared to traditional methods.

This scientific work and its results were presented to the general public and specialists at the prestigious international KDD 2026 conference, receiving high praise from the international scientific community.

In the future, the research team plans to further expand the capabilities of this technology. In particular, it is planned to take user feedback into account more accurately and apply it in wider industrial tasks.