Graph Contrastive Learning on Multi-label Classification for Recommendations

TL;DR

This work addresses accurate link prediction in bipartite recommendation graphs with multi-label outcomes. It introduces MCGCL, a graph-contrastive learning framework that jointly trains a holistic main task on the full user-item graph and a subtask over homogeneous subgraphs, using edge-perturbation augmentations and entropy-based hard-sample mining. The approach combines label-specific encoders, attention-based fusion, and cross-view projections, achieving superior results over end-to-end, signed-GNN, and other contrastive baselines on six Amazon datasets for both multi-label and binary predictions. The findings demonstrate that incorporating sub-view learning and contrastive objectives yields robust representations, offering practical improvements for recommendation systems, especially in sparse or complex multi-label settings.

Abstract

In business analysis, providing effective recommendations is essential for enhancing company profits. The utilization of graph-based structures, such as bipartite graphs, has gained popularity for their ability to analyze complex data relationships. Link prediction is crucial for recommending specific items to users. Traditional methods in this area often involve identifying patterns in the graph structure or using representational techniques like graph neural networks (GNNs). However, these approaches encounter difficulties as the volume of data increases. To address these challenges, we propose a model called Graph Contrastive Learning for Multi-label Classification (MCGCL). MCGCL leverages contrastive learning to enhance recommendation effectiveness. The model incorporates two training stages: a main task and a subtask. The main task is holistic user-item graph learning to capture user-item relationships. The homogeneous user-user (item-item) subgraph is constructed to capture user-user and item-item relationships in the subtask. We assessed the performance using real-world datasets from Amazon Reviews in multi-label classification tasks. Comparative experiments with state-of-the-art methods confirm the effectiveness of MCGCL, highlighting its potential for improving recommendation systems.

Figures (10)

• Figure 1: Illustrative scenarios of link prediction.
• Figure 2: Example of link prediction based on user ratings.
• Figure 3: Illustrative diagram of a multi-label bipartite graph.
• Figure 4: Comparative analysis using the same encoder for $G_{1}, G^{'}_{1}$, and using different encoders for $G_{1}, G_{2}$.
• Figure 5: Overview of the training process.

Theorems & Definitions (4)

• definition 1: Problem definition
• definition 2: Graph encoder layer
• definition 3: Contrastive loss function
• definition 4: Homogeneous graph and holistic graph