A Generative Re-ranking Model for List-level Multi-objective Optimization at Taobao

2025-05-12

AI Summary

This paper addresses the list-level multi-objective optimization challenge in e-commerce recommendation, where conventional item-level methods neglect dynamic user intent and inter-item interactions, while existing re-ranking models struggle to jointly optimize efficiency, diversity, and multiple objectives. We propose SORT-Gen, an end-to-end generative re-ranking model. Its key contributions are: (1) a novel Sequentially Ordered Regression Transformer that models the joint distribution of click-through rate (CTR), conversion rate (CVR), and gross merchandise volume (GMV) over variable-length sublists; and (2) a mask-driven fast generation algorithm integrating multi-objective candidate queues, efficient item selection, and explicit diversity control. SORT-Gen supports online real-time inference and has been deployed in core scenarios of the Taobao mobile app. Online A/B tests demonstrate significant improvements of +4.13% in CTR and +8.10% in GMV, validating the effectiveness and practicality of the list-level multi-objective generative paradigm.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Search and Optimization: Algorithm Configuration
• Data Mining & Knowledge Management: Recommender Systems

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions

Abstract

E-commerce recommendation systems aim to generate ordered lists of items for customers, optimizing multiple business objectives, such as clicks, conversions and Gross Merchandise Volume (GMV). Traditional multi-objective optimization methods like formulas or Learning-to-rank (LTR) models take effect at item-level, neglecting dynamic user intent and contextual item interactions. List-level multi-objective optimization in the re-ranking stage can overcome this limitation, but most current re-ranking models focus more on accuracy improvement with context. In addition, re-ranking is faced with the challenges of time complexity and diversity. In light of this, we propose a novel end-to-end generative re-ranking model named Sequential Ordered Regression Transformer-Generator (SORT-Gen) for the less-studied list-level multi-objective optimization problem. Specifically, SORT-Gen is divided into two parts: 1)Sequential Ordered Regression Transformer innovatively uses Transformer and ordered regression to accurately estimate multi-objective values for variable-length sub-lists. 2)Mask-Driven Fast Generation Algorithm combines multi-objective candidate queues, efficient item selection and diversity mechanism into model inference, providing a fast online list generation method. Comprehensive online experiments demonstrate that SORT-Gen brings +4.13% CLCK and +8.10% GMV for Baiyibutie, a notable Mini-app of Taobao. Currently, SORT-Gen has been successfully deployed in multiple scenarios of Taobao App, serving for a vast number of users.

Problem

• Optimizes e-commerce recommendations for multiple business objectives
• Addresses item-level limitations by focusing on list-level optimization
• Solves time complexity and diversity challenges in re-ranking

Innovation

• Generative re-ranking model for multi-objective optimization
• Transformer and ordered regression for accurate estimation
• Fast generation algorithm with diversity mechanism