USD: A User-Intent-Driven Sampling and Dual-Debiasing Framework for Large-Scale Homepage Recommendations

TL;DR

This paper introduces a unified framework for large-scale homepage recommendation that effectively addresses exposure bias and pseudo-positive samples, significantly improving user click-through rates through user intent-aware sampling and dual-debiasing.

Contribution

It proposes a novel user intent-aware negative sampling and dual-debiasing framework specifically designed for homepage recommendation systems.

Findings

Achieved 35.4% and 14.5% improvements in user click-through rates.

Effectively filters invalid exposure samples and corrects exposure and click biases.

Demonstrated effectiveness through extensive online experiments on Taobao.

Abstract

Large-scale homepage recommendations face critical challenges from pseudo-negative samples caused by exposure bias, where non-clicks may indicate inattention rather than disinterest. Existing work lacks thorough analysis of invalid exposures and typically addresses isolated aspects (e.g., sampling strategies), overlooking the critical impact of pseudo-positive samples - such as homepage clicks merely to visit marketing portals. We propose a unified framework for large-scale homepage recommendation sampling and debiasing. Our framework consists of two key components: (1) a user intent-aware negative sampling module to filter invalid exposure samples, and (2) an intent-driven dual-debiasing module that jointly corrects exposure bias and click bias. Extensive online experiments on Taobao demonstrate the efficacy of our framework, achieving significant improvements in user click-through rates (UCTR) by 35.4% and 14.5% in two variants of the marketing block on the Taobao homepage, Baiyibutie and Taobaomiaosha.