Каскадна система оцінювання для покращення рекомендацій настільних ігор

Abstract

The paper proposes an approach to improving the recommendation system of board games by introducing a cascade evaluation system. The basic principles of building such a system and its advantages in comparison with traditional approaches are considered. The role of the Two Tower model as a component in the cascade architecture is analysed. It is shown that the proposed approach can significantly improve the quality of recommendations while maintaining high system performance.