Self-EvolveRec: Self-Evolving Recommender Systems with LLM-based Directional Feedback

Researchers propose Self-EvolveRec, a novel framework that uses LLMs to create self-evolving recommender systems. It moves beyond traditional Neural Architecture Search by establishing a directional feedback loop with qualitative critiques from a User Simulator and quantitative verification from a Model Diagnosis Tool. Experiments show it significantly outperforms state-of-the-art NAS and LLM-driven code evolution baselines in both recommendation performance and user satisfaction. The code is publicly available.