US Patent 11995665 Reducing sample selection bias in a machine learning-based recommender system

The present disclosure relates to improving recommendations for small shops on an ecommerce platform while maintaining accuracy for larger shops. The improvement is achieved by retraining a machine-learning recommendation model to reduce sample selection bias using a meta-learning process. The retraining process comprises identifying a sample subset of shops on the ecommerce platform, and then creating shop-specific versions of the recommendation model for each of the shops in the subset. A global parameter adjustment is calculated for the global model based on minimizing losses associated with the shop-specific models and increasing the probability of items being recommended from small shops. The latter is achieved by introducing regularizer terms for small shops during the meta-learning process. The regularizer terms serve to increase the probability that an item from a small shop will be recommended, thereby countering the sample selection bias faced by small-shop items.