Optimizing Novelty of Top-k Recommendations using Large Language Models and Reinforcement Learning