Self-EvolveRec Framework Uses LLM Qualitative Feedback for Recommender Evolution
July 10, 2026
Self-EvolveRec moves beyond scalar metrics like NDCG by using a User Simulator and Model Diagnosis Tool to provide qualitative, directional feedback for code evolution. The framework employs a Diagnosis Tool-Model Co-Evolution strategy to dynamically adapt evaluation criteria as recommendation architectures change.
HOW THIS AFFECTS YOU
●
builderThis offers a path toward self-improving recommendation engines that learn from qualitative user critiques.
●
researcherYou can move beyond fixed search spaces and scalar optimization in automated architecture design.