Abstract
Post-click conversion rate (CVR) estimation is a fundamental task in developing effective recommender systems, yet it faces challenges from data sparsity and sample selection bias. To handle both challenges, employ entire space multitask models decompose the user behavior track into a sequence of exposure → click → conversion, to foster construct auxiliary learning tasks for CVR estimation. However, these methods suffer from two significant defects: (1) intrinsic estimation bias (IEB), where CVR estimates are in overestimated; (2) false independence prior (FIP), where the causal relationship between clicks and subsequent conversions might be overlooked. To overcome these limitations, we develop a model-agnostic framework, namely Entire Space Counterfactual Multitask Model (ESCM2), which incorporates a counterfactual risk minimizer within the ESMM framework to regularize CVR estimation. Experiments conducted on large-scale industrial recommendation datasets and an online industrial recommendation service demonstrate ESCM 2 effectively mitigates IEB and FIP defects and substantially enhances recommendation performance.
Funding Information
  • National Natural Science Foundation of China (623B2002, 62073288, 12075212, 12105246)

This publication has 42 references indexed in Scilit: