Researcher, Huawei Noah's Ark Lab
2 papers at NeurIPS 2025
LLM-as-RS paradigm is not always good: we benchmark it with traditional recommendation models