A Hybrid Cross-Stage Coordination Pre-ranking Model for Online Recommendation Systems

TL;DR

This paper introduces HCCP, a hybrid pre-ranking model that integrates upstream and downstream information to improve recommendation accuracy and mitigate sample selection bias in large-scale systems.

Contribution

The paper proposes a novel hybrid cross-stage coordination pre-ranking model with hybrid sample construction and hybrid objective optimization, enhancing performance and addressing sample bias.

Findings

HCCP outperforms state-of-the-art methods in offline and online tests.

Achieves up to 14.9% UCVR and 1.3% UCTR improvements in JD E-commerce.

Effectively mitigates sample selection bias and Matthew effect.

Abstract

Large-scale recommendation systems often adopt cascading architecture consisting of retrieval, pre-ranking, ranking, and re-ranking stages. With strict latency requirements, pre-ranking utilizes lightweight models to perform a preliminary selection from massive retrieved candidates. However, recent works focus solely on improving consistency with ranking, relying exclusively on downstream stages. Since downstream input is derived from the pre-ranking output, they will exacerbate the sample selection bias (SSB) issue and Matthew effect, leading to sub-optimal results. To address the limitation, we propose a novel Hybrid Cross-Stage Coordination Pre-ranking model (HCCP) to integrate information from upstream (retrieval) and downstream (ranking, re-ranking) stages. Specifically, cross-stage coordination refers to the pre-ranking's adaptability to the entire stream and the role of serving as a more effective bridge between upstream and downstream. HCCP consists of Hybrid Sample Construction and Hybrid Objective Optimization. Hybrid sample construction captures multi-level unexposed data from the entire stream and rearranges them to become the optimal guiding "ground truth" for pre-ranking learning. Hybrid objective optimization contains the joint optimization of consistency and long-tail precision through our proposed Margin InfoNCE loss. It is specifically designed to learn from such hybrid unexposed samples, improving the overall performance and mitigating the SSB issue. The appendix describes a proof of the efficacy of the proposed loss in selecting potential positives. Extensive offline and online experiments indicate that HCCP outperforms SOTA methods by improving cross-stage coordination. It contributes up to 14.9% UCVR and 1.3% UCTR in the JD E-commerce recommendation system. Concerning code privacy, we provide a pseudocode for reference.