bpr

Definition

Bayesian Personalized Ranking, a standard objective for learning recommendations from implicit feedback such as clicks and purchases. Rather than predicting a rating, it optimizes pairwise order: for each user, an item they interacted with should score above one they did not. That reframing matters because implicit logs contain no explicit negatives — non-interaction is ambiguous — and pairwise ranking sidesteps having to interpret it as dislike.

Explainers using this term

• Recommenders and Embeddings — Same Math as RAG, Different GoalBPR: Bayesian Personalized Ranking from Implicit Feedback