How to solve the problem of frequent recommendation of popular products in retail?

The problem of frequent recommendation of popular products in retail is related to the fact that popular items often dominate recommendations, reducing diversity and personalization. To address this issue, several approaches can be applied:

• Introducing diversity: use algorithms that consider not only popularity but also the variety of products, for example, by limiting the repetition of popular items in recommendations.

• Personalization: build recommendations based on the individual preferences of the user, using collaborative filtering or content-based models.

• Popularity regularization: reduce the weight of popular products in the recommendation model to give a chance to less popular items.

• Using ranking algorithms that consider business goals: for example, optimize recommendations to increase the average check or promote new products.

Example: you can implement an algorithm that first forms a list of candidates considering the user's preferences, and then applies a diversity filter, excluding overly frequently recommended popular products.