Heterogeneous Graph Masked Contrastive Learning for Robust Recommendation

TL;DR

This paper introduces Masked Contrastive Learning (MCL), a novel approach that enhances the robustness of heterogeneous graph neural network-based recommendations by reducing noise sensitivity through graph augmentation and contrastive learning.

Contribution

The paper proposes MCL, a new model that employs random masking and contrastive learning on heterogeneous information networks to improve recommendation robustness against noisy graphs.

Findings

MCL outperforms existing methods on three real-world datasets.

Graph augmentation via masking reduces noise impact on embeddings.

Contrastive learning captures both local and high-order structures effectively.

Abstract

Heterogeneous graph neural networks (HGNNs) have demonstrated their superiority in exploiting auxiliary information for recommendation tasks. However, graphs constructed using meta-paths in HGNNs are usually too dense and contain a large number of noise edges. The propagation mechanism of HGNNs propagates even small amounts of noise in a graph to distant neighboring nodes, thereby affecting numerous node embeddings. To address this limitation, we introduce a novel model, named Masked Contrastive Learning (MCL), to enhance recommendation robustness to noise. MCL employs a random masking strategy to augment the graph via meta-paths, reducing node sensitivity to specific neighbors and bolstering embedding robustness. Furthermore, MCL employs contrastive cross-view on a Heterogeneous Information Network (HIN) from two perspectives: one-hop neighbors and meta-path neighbors. This approach acquires embeddings capturing both local and high-order structures simultaneously for recommendation. Empirical evaluations on three real-world datasets confirm the superiority of our approach over existing recommendation methods.