Causal Negative Sampling via Diffusion Model for Out-of-Distribution Recommendation

TL;DR

This paper introduces CNSDiff, a diffusion-based negative sampling method that reduces bias from confounders in recommendation systems, improving out-of-distribution generalization.

Contribution

It proposes a novel causal negative sampling approach using diffusion models and causal regularization to enhance OOD recommendation robustness.

Findings

CNSDiff outperforms baselines by 13.96% on average across metrics.

The method effectively mitigates bias from confounders in negative sampling.

Experiments demonstrate improved robustness under distribution shifts.

Abstract

Heuristic negative sampling enhances recommendation performance by selecting negative samples of varying hardness levels from predefined candidate pools to guide the model toward learning more accurate decision boundaries. However, our empirical and theoretical analyses reveal that unobserved environmental confounders (e.g., exposure or popularity biases) in candidate pools may cause heuristic sampling methods to introduce false hard negatives (FHNS). These misleading samples can encourage the model to learn spurious correlations induced by such confounders, ultimately compromising its generalization ability under distribution shifts. To address this issue, we propose a novel method named Causal Negative Sampling via Diffusion (CNSDiff). By synthesizing negative samples in the latent space via a conditional diffusion process, CNSDiff avoids the bias introduced by predefined candidate pools and thus reduces the likelihood of generating FHNS. Moreover, it incorporates a causal regularization term to explicitly mitigate the influence of environmental confounders during the negative sampling process, leading to robust negatives that promote out-of-distribution (OOD) generalization. Comprehensive experiments under four representative distribution shift scenarios demonstrate that CNSDiff achieves an average improvement of 13.96% across all evaluation metrics compared to state-of-the-art baselines, verifying its effectiveness and robustness in OOD recommendation tasks.