Trading Engagement for Sustainability: Carbon-Aware Re-ranking for E-commerce Recommendations

In three linesStudy on e-commerce recommender systems incorporating product carbon footprint. Researchers estimate missing PCF via semantic search and LLM prompting, then apply post-hoc re-ranking on BPR, NeuMF, and LightGCN. On Amazon Reviews (3 categories), substantial carbon reductions achievable with minimal engagement cost.

Summary generated by Claude — human-verified