RIA: A Ranking-Infused Approach for Optimized listwise CTR Prediction

2025-11-26

AI Summary

In existing recommender systems, decoupling ranking from re-ranking leads to combinatorial sparsity and limited representational capacity for list-wise evaluation models under strict latency constraints. To address this, we propose RIA—the first end-to-end CTR prediction framework that jointly optimizes ranking and re-ranking, unifying point-wise and list-wise signal modeling. Key innovations include a dual Transformer architecture for user-candidate interaction, context-aware user history modeling (CUHT), hierarchical list modeling via Multi-HSTU, and an embedding caching mechanism—collectively balancing accuracy and latency. Extensive experiments on both public and industrial datasets demonstrate significant improvements over state-of-the-art methods. Online A/B testing shows a 1.69% lift in CTR and a 4.54% increase in CPM, validating RIA’s practical efficacy in production environments.

Technology Category

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

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
• Responsible Web: Data and user privacy-enhancing technologies for the Web

Abstract

Reranking improves recommendation quality by modeling item interactions. However, existing methods often decouple ranking and reranking, leading to weak listwise evaluation models that suffer from combinatorial sparsity and limited representational power under strict latency constraints. In this paper, we propose RIA (Ranking-Infused Architecture), a unified, end-to-end framework that seamlessly integrates pointwise and listwise evaluation. RIA introduces four key components: (1) the User and Candidate DualTransformer (UCDT) for fine-grained user-item-context modeling; (2) the Context-aware User History and Target (CUHT) module for position-sensitive preference learning; (3) the Listwise Multi-HSTU (LMH) module to capture hierarchical item dependencies; and (4) the Embedding Cache (EC) module to bridge efficiency and effectiveness during inference. By sharing representations across ranking and reranking, RIA enables rich contextual knowledge transfer while maintaining low latency. Extensive experiments show that RIA outperforms state-of-the-art models on both public and industrial datasets, achieving significant gains in AUC and LogLoss. Deployed in Meituan advertising system, RIA yields a +1.69% improvement in Click-Through Rate (CTR) and a +4.54% increase in Cost Per Mille (CPM) in online A/B tests.

Problem

• Optimizing listwise CTR prediction by integrating ranking and reranking stages
• Addressing combinatorial sparsity and limited representational power in recommendations
• Maintaining low latency while capturing rich item interactions

Innovation

• Integrates pointwise and listwise evaluation end-to-end
• Uses DualTransformer for fine-grained user-item-context modeling
• Employs Embedding Cache module for efficiency and effectiveness