Multi-view debiasing representation learning for recommender systems

doi:10.1016/j.ipm.2025.104429

Multi-view debiasing representation learning for recommender systems

2025-10-07

0

PRE

AI

DOI：10.1016/j.ipm.2025.104429

Original

Original request for help

Save

Abstract

En 中文

• Proposed a novel Multi-View Identifiable Debiased Learning (MViDL) method to integrate diverse data views for complex real-world scenarios. • Theoretically analysed the identifiability of latent representations learned by MViDL. • Achieved state-of-the-art performance in mitigating confounding bias on three real-world datasets.

Multi-view debiasing representation learning for recommender systems

Abstract

Journal

Organization