Dynamic Graph Recommendation via Sparse Augmentation and Singular Adaptation

TL;DR

GraphSASA is a novel fine-tuning method for dynamic graph recommendation systems that enhances efficiency and performance by using test-time augmentation and singular value decomposition to adapt pre-trained GNNs.

Contribution

It introduces a new fine-tuning approach combining test-time augmentation and SVD to reduce computational costs and improve adaptability in dynamic recommendation models.

Findings

Achieves state-of-the-art results on three large-scale datasets.

Reduces fine-tuning parameters while maintaining high performance.

Enhances node representations through hierarchical graph augmentation.

Abstract

Dynamic recommendation, focusing on modeling user preference from historical interactions and providing recommendations on current time, plays a key role in many personalized services. Recent works show that pre-trained dynamic graph neural networks (GNNs) can achieve excellent performance. However, existing methods by fine-tuning node representations at large scales demand significant computational resources. Additionally, the long-tail distribution of degrees leads to insufficient representations for nodes with sparse interactions, posing challenges for efficient fine-tuning. To address these issues, we introduce GraphSASA, a novel method for efficient fine-tuning in dynamic recommendation systems. GraphSASA employs test-time augmentation by leveraging the similarity of node representation distributions during hierarchical graph aggregation, which enhances node representations. Then it applies singular value decomposition, freezing the original vector matrix while focusing fine-tuning on the derived singular value matrices, which reduces the parameter burden of fine-tuning and improves the fine-tuning adaptability. Experimental results demonstrate that our method achieves state-of-the-art performance on three large-scale datasets.