Pith. sign in

Mitigating Dual Latent Confounding Biases in Recommender Systems

1 Pith paper cite this work. Polarity classification is still indexing.

1 Pith paper citing it
abstract

Recommender systems are extensively utilised across various areas to predict user preferences for personalised experiences and enhanced user engagement and satisfaction. Traditional recommender systems, however, are complicated by confounding bias, particularly in the presence of latent confounders that affect both item exposure and user feedback. Existing debiasing methods often fail to capture the complex interactions caused by latent confounders in interaction data, especially when dual latent confounders affect both the user and item sides. To address this, we propose a novel debiasing method that jointly integrates the Instrumental Variables (IV) approach and identifiable Variational Auto-Encoder (iVAE) for Debiased representation learning in Recommendation systems, referred to as IViDR. Specifically, IViDR leverages the embeddings of user features as IVs to address confounding bias caused by latent confounders between items and user feedback, and reconstructs the embedding of items to obtain debiased interaction data. Moreover, IViDR employs an Identifiable Variational Auto-Encoder (iVAE) to infer identifiable representations of latent confounders between item exposure and user feedback from both the original and debiased interaction data. Additionally, we provide theoretical analyses of the soundness of using IV and the identifiability of the latent representations. Extensive experiments on both synthetic and real-world datasets demonstrate that IViDR outperforms state-of-the-art models in reducing bias and providing reliable recommendations.

citation-role summary

background 1

citation-polarity summary

fields

cs.IR 1

years

2025 1

verdicts

CONDITIONAL 1

roles

background 1

polarities

unclear 1

representative citing papers

Generating Negative Samples for Multi-Modal Recommendation

cs.IR · 2025-01-25 · conditional · novelty 6.0

NegGen masks and replaces key attributes of items via a multi-modal LLM, generating hard negative descriptions, and uses a contrastive causal module to improve multi-modal recommendation across four Amazon datasets.

citing papers explorer

Showing 1 of 1 citing paper.

  • Generating Negative Samples for Multi-Modal Recommendation cs.IR · 2025-01-25 · conditional · none · ref 7 · internal anchor

    NegGen masks and replaces key attributes of items via a multi-modal LLM, generating hard negative descriptions, and uses a contrastive causal module to improve multi-modal recommendation across four Amazon datasets.