ACT: Automated Constraint Targeting for Multi-Objective Recommender Systems
This addresses the challenge of maintaining consistent user experience and platform ecosystem in recommender systems, though it appears incremental as it builds on existing constraint satisfaction methods.
The paper tackles the problem of enforcing secondary objective guardrails in multi-objective recommender systems by introducing the Automated Constraint Targeting (ACT) framework, which automatically finds minimal hyperparameter changes to satisfy these constraints and was deployed in a large-scale production environment.
Recommender systems often must maximize a primary objective while ensuring secondary ones satisfy minimum thresholds, or "guardrails." This is critical for maintaining a consistent user experience and platform ecosystem, but enforcing these guardrails despite orthogonal system changes is challenging and often requires manual hyperparameter tuning. We introduce the Automated Constraint Targeting (ACT) framework, which automatically finds the minimal set of hyperparameter changes needed to satisfy these guardrails. ACT uses an offline pairwise evaluation on unbiased data to find solutions and continuously retrains to adapt to system and user behavior changes. We empirically demonstrate its efficacy and describe its deployment in a large-scale production environment.