Personalized LLM Response Generation System for Reducing Household Food Waste

Food waste is a critical issue impacting many aspects of environmental sustainability. Importantly, approximately 60 percent of food waste results from households. In this paper, we develop the third version of SmartFridge, an app to reduce household food waste. SmartFridge 3.0 features our proposed PLLMRGS, a Personalized LLM Response Generation System that considers both the user’s long-term preferences and short-term preferences. Long-term and short-term preferences are extracted from user preference data using a Heavy Hitters algorithm and a Top-N algorithm, respectively, and combined into a cohesive LLM prompt. Our experiments show that PLLMRGS outperforms GPT-4o mini in recipe generation, demonstrating more personalized and satisfactory results. The benefits of PLLMRGS boost the viability and desirability of SmartFridge 3.0, creating broader impact in mitigating house-hold food waste.