Trading Engagement for Sustainability: Carbon-Aware Re-ranking for E-commerce Recommendations
持続可能性のためのエンゲージメントのトレードオフ：eコマースレコメンデーションのための炭素認識再ランキング (AI 翻訳)
Noah Lund Syrdal, Anders Vestrum, Jörgen Bergh
🤖 gxceed AI 要約
日本語
本論文は、eコマースにおいて製品カーボンフットプリント（PCF）を推定し、炭素認識の再ランキング戦略を提案する。Carbon Catalogueからの転移学習によりPCFを推定し、BPR、NeuMF、LightGCNモデル上でエンゲージメントと炭素排出のトレードオフを実現。Amazonレビューデータセットでの評価により、最小限のエンゲージメント損失で大きな炭素削減が可能であることを示した。
English
This paper proposes a carbon-aware re-ranking strategy for e-commerce recommendations by estimating product carbon footprints via transfer learning from the Carbon Catalogue. Using three recommendation models (BPR, NeuMF, LightGCN) on Amazon Reviews, it demonstrates that substantial carbon reductions can be achieved with minimal engagement loss.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本のeコマースプラットフォームにおいて、製品カーボンフットプリント表示や持続可能な購買行動促進に活用可能。SSBJや有報でのカーボンフットプリント開示の実務にも示唆を与える。
In the global GX context
This work contributes to global carbon accounting in retail by providing a scalable method for PCF estimation and integration into recommendation systems, aligning with emerging PCF disclosure requirements and sustainability goals.
👥 読者別の含意
🔬研究者:Provides a novel methodology for carbon footprint estimation and its integration into recommender systems, with empirical trade-off analysis.
🏢実務担当者:E-commerce platforms can adopt the proposed re-ranking strategy to reduce carbon footprint of product recommendations without significantly affecting user engagement.
📄 Abstract（原文）
E-commerce recommender systems strongly influence which products users consider and purchase, yet sustainability signals such as Product Carbon Footprint (PCF) are almost never available at catalog scale. We study carbon-aware product recommendation in the realistic setting where PCF labels are missing for most items and must be inferred. We first estimate product-level carbon footprints via a retrieval-augmented PCF estimation pipeline that transfers supervision from the Carbon Catalogue, a small set of life-cycle-assessed products, to a large unlabeled e-commerce catalog using semantic similarity search, few-shot LLM prompting, and a nearest-neighbour fallback. We then apply a carbon-aware post-hoc re-ranking strategy on top of relevance scores produced by three established recommendation models: BPR, NeuMF, and LightGCN. The method trades off predicted user-item engagement against estimated carbon footprint through a single tunable parameter, lambda. In this offline study, engagement is operationalized through Amazon review interactions, which serve as implicit feedback and as a proxy for user interest or purchase behavior. We evaluate the framework on the Amazon Reviews dataset across three product categories: Home and Kitchen, Sports and Outdoors, and Electronics. By sweeping lambda, we construct Pareto frontiers that characterize the achievable engagement and carbon trade-off for each model and category. Substantial carbon reductions are achievable at minimal engagement cost across all models and categories. However, the available carbon headroom varies by model and category, underscoring the importance of model choice and domain context.
🔗 Provenance — このレコードを発見したソース
- openalex https://doi.org/10.48550/arxiv.2606.04550first seen 2026-06-06 04:52:37
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録（無料・3キーワードまで）。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。