Enhancing Graph Contrastive Learning with Reliable and Informative Augmentation for Recommendation

TL;DR

This work tackles the challenge of data sparsity in graph neural network–based recommender systems by uncovering that existing graph contrastive learning view perturbations can erode collaborative information. It introduces CoGCL, which uses an end-to-end multi-level vector quantizer to map users and items to discrete codes, and leverages these codes to construct reliable and informative contrastive views via virtual neighbor augmentation and semantic relevance sampling. A triple-view graph contrastive learning objective then aligns two augmented views and semantically related users/items, jointly optimizing with the standard BPR loss. Empirical results on four public datasets show that CoGCL consistently outperforms strong baselines, especially under sparse conditions, and ablations confirm the necessity of each component. The method offers a scalable, information-rich alternative for enhancing graph CL in recommendation systems.

Abstract

Graph neural network(GNN) has been a powerful approach in collaborative filtering(CF) due to its ability to model high-order user-item relationships. Recently, to alleviate the data sparsity and enhance representation learning, many efforts have been conducted to integrate contrastive learning(CL) with GNNs. Despite the promising improvements, the contrastive view generation based on structure and representation perturbations in existing methods potentially disrupts the collaborative information in contrastive views, resulting in limited effectiveness of positive alignment. To overcome this issue, we propose CoGCL, a novel framework that aims to enhance graph contrastive learning by constructing contrastive views with stronger collaborative information via discrete codes. The core idea is to map users and items into discrete codes rich in collaborative information for reliable and informative contrastive view generation. To this end, we initially introduce a multi-level vector quantizer in an end-to-end manner to quantize user and item representations into discrete codes. Based on these discrete codes, we enhance the collaborative information of contrastive views by considering neighborhood structure and semantic relevance respectively. For neighborhood structure, we propose virtual neighbor augmentation by treating discrete codes as virtual neighbors, which expands an observed user-item interaction into multiple edges involving discrete codes. Regarding semantic relevance, we identify similar users/items based on shared discrete codes and interaction targets to generate the semantically relevant view. Through these strategies, we construct contrastive views with stronger collaborative information and develop a triple-view graph contrastive learning approach. Extensive experiments on four public datasets demonstrate the effectiveness of our proposed approach.

Figures (9)

• Figure 1: Comparison of current graph CL-based methods (e.g., SGL DBLP:conf/sigir/WuWF0CLX21, SimGCL DBLP:conf/sigir/YuY00CN22) that disrupt collaborative information within contrastive views and the proposed approach that enhances collaborative information.
• Figure 2: Performance comparison of different graph CL-based methods with their variants.
• Figure 3: The overall framework of our CoGCL, which enhances graph CL by constructing contrastive views that imply stronger collaborative information via discrete codes.
• Figure 4: Ablation study of data augmentation methods.
• Figure 5: Performance comparison of different discrete code learning methods.