ChorusCVR: Chorus Supervision for Entire Space Post-Click Conversion Rate Modeling

TL;DR

This work tackles sample selection bias in post-click CVR estimation by training CVR in the entire exposure space rather than solely in the click space. It introduces ChorusCVR, which consists of a Negative sample Discrimination Module (NDM) producing discriminative soft CTunCVR labels and a Soft Alignment Module (SAM) that using these labels to supervise CVR learning with robust alignment across spaces. The CTunCVR objective and IPW-based stabilization enable robust discrimination between factual and ambiguous negatives, while SAM provides mutual supervision between CVR and unCVR predictions, reducing bias in un-clicked samples. Offline experiments on Ali-CCP and Kuaishou show consistent gains over state-of-the-art baselines in CVR-AUC and CTCVR-AUC, and online A/B tests confirm practical improvements in CVR, orders, and DAC, demonstrating the method’s effectiveness in production environments.

Abstract

Post-click conversion rate (CVR) estimation is a vital task in many recommender systems of revenue businesses, e.g., e-commerce and advertising. In a perspective of sample, a typical CVR positive sample usually goes through a funnel of exposure to click to conversion. For lack of post-event labels for un-clicked samples, CVR learning task commonly only utilizes clicked samples, rather than all exposed samples as for click-through rate (CTR) learning task. However, during online inference, CVR and CTR are estimated on the same assumed exposure space, which leads to a inconsistency of sample space between training and inference, i.e., sample selection bias (SSB). To alleviate SSB, previous wisdom proposes to design novel auxiliary tasks to enable the CVR learning on un-click training samples, such as CTCVR and counterfactual CVR, etc. Although alleviating SSB to some extent, none of them pay attention to the discrimination between ambiguous negative samples (un-clicked) and factual negative samples (clicked but un-converted) during modelling, which makes CVR model lacks robustness. To full this gap, we propose a novel ChorusCVR model to realize debiased CVR learning in entire-space.

Figures (3)

• Figure 1: A conceptual comparison between our proposed ChorusCVR and existing CVR models on the perspective of the discrimination spaces of soft labels.
• Figure 2: Systematic overview of our Chorus CVR model.
• Figure 3: The PCOC analysis.