A Cascaded Generative Approach for e-Commerce Recommendations

Moein Hasani, Hamidreza Shahidi, Trace Levinson, Yuan Zhong, Guanghua Shu, Vinesh Gudla, Tejaswi Tenneti

arXiv:2605.11118 · 2026-05-13 공개 · arXiv · PDF

e-commerce recommendation-systems placement-generation keyword-generation teacher-student-finetuning content-evaluation ranking-models llm-ablation

Abstract

Personalized storefronts in large e-commerce marketplaces are often assembled from many independent components: static themes per page section ("placement"), retrieval systems to fetch eligible products per placement, and pointwise rankers to order content. While effective in optimizing for aggregate preferences, this paradigm is rigid and can limit personalization and semantic cohesion across the page. This makes it poorly suited to support dynamic objectives and merchandising requirements over time. To address this, we introduce a cascaded merchandising framework that decomposes storefront construction into two generative tasks: (i) placement-level theme generation and (ii) constrained keyword generation per placement to power product retrieval. Teacher-student fine-tuning is leveraged to improve scalability of this framework under production latency and cost constraints. Fine-tuned model ablations are shown to approach closed-weight LLM performance. We further contribute frameworks for AI-driven content evaluation and quality filtering, enabling safe and automated deployment of dynamic content at scale. Generative output is fused with traditional ranking models to preserve hybrid infrastructure. In online experiments, this framework yields an estimated +2.7% lift in cart adds per page view over a strong baseline.

한국어 요약

📋 한 줄 요약

**[추천 시스템 / 생성형 AI]** 전자상거래 스토어프론트 구성을 placement-level 테마 생성과 키워드 생성 두 단계 생성 태스크로 분해하는 cascaded 머천다이징 프레임워크 제안.

🎯 핵심 기여도

💡 핵심 아이디어

전자상거래 페이지를 "정적 슬롯에 제품을 채워 넣는" 방식이 아니라 "테마와 키워드가 동적으로 생성되어 페이지 전체의 의미적 응집성을 만들어내는" 생성 태스크로 재정의. 생성 출력은 기존 랭킹 모델과 fusion되어 하이브리드 인프라를 보존한다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 검색·랭킹 중심 e-commerce 스택에 생성형 AI를 안전하게 통합하는 production-grade 청사진. **한계**: 평가가 한 마켓플레이스 도메인 중심이며, 신규 카테고리나 콜드스타트 시나리오 일반화는 추가 검증이 필요.

🚀 실용적 활용