Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Multi-behavior Recommendation with SVD Graph Neural Networks

Shengxi Fu, Qianqian Ren, Xingfeng Lv, Jinbao Li

TL;DR

MB-SVD tackles multi-behavior recommendation by combining per-behavior GCN embeddings with an SVD-based augmentation of the user-item graph to capture global collaborative signals. It introduces a behavior-aware aggregation to weight different behaviors and a simplified InfoNCE-based contrastive objective that contrasts SVD-augmented views with original representations, all optimized jointly with a BPR-style loss. The approach yields superior performance on three real-world datasets, particularly Taobao and Yelp, and demonstrates robustness to data sparsity through the SVD augmentation and contrastive learning components. Overall, MB-SVD advances multi-behavior recommendation by effectively integrating local behavior signals with global structure while mitigating over-smoothing and noise.

Abstract

Graph Neural Networks (GNNs) have been extensively employed in the field of recommendation systems, offering users personalized recommendations and yielding remarkable outcomes. Recently, GNNs incorporating contrastive learning have demonstrated promising performance in handling the sparse data problem of recommendation systems. However, existing contrastive learning methods still have limitations in resisting noise interference, especially for multi-behavior recommendation. To mitigate the aforementioned issues, this paper proposes a GNN-based multi-behavior recommendation model called MB-SVD that utilizes Singular Value Decomposition (SVD) graphs to enhance model performance. In particular, MB-SVD considers user preferences across different behaviors, improving recommendation effectiveness. First, MB-SVD integrates the representation of users and items under different behaviors with learnable weight scores, which efficiently considers the influence of different behaviors. Then, MB-SVD generates augmented graph representation with global collaborative relations. Next, we simplify the contrastive learning framework by directly contrasting original representation with the enhanced representation using the InfoNCE loss. Through extensive experimentation, the remarkable performance of our proposed MB-SVD approach in multi-behavior recommendation endeavors across diverse real-world datasets is exhibited.

Multi-behavior Recommendation with SVD Graph Neural Networks

TL;DR

MB-SVD tackles multi-behavior recommendation by combining per-behavior GCN embeddings with an SVD-based augmentation of the user-item graph to capture global collaborative signals. It introduces a behavior-aware aggregation to weight different behaviors and a simplified InfoNCE-based contrastive objective that contrasts SVD-augmented views with original representations, all optimized jointly with a BPR-style loss. The approach yields superior performance on three real-world datasets, particularly Taobao and Yelp, and demonstrates robustness to data sparsity through the SVD augmentation and contrastive learning components. Overall, MB-SVD advances multi-behavior recommendation by effectively integrating local behavior signals with global structure while mitigating over-smoothing and noise.

Abstract

Graph Neural Networks (GNNs) have been extensively employed in the field of recommendation systems, offering users personalized recommendations and yielding remarkable outcomes. Recently, GNNs incorporating contrastive learning have demonstrated promising performance in handling the sparse data problem of recommendation systems. However, existing contrastive learning methods still have limitations in resisting noise interference, especially for multi-behavior recommendation. To mitigate the aforementioned issues, this paper proposes a GNN-based multi-behavior recommendation model called MB-SVD that utilizes Singular Value Decomposition (SVD) graphs to enhance model performance. In particular, MB-SVD considers user preferences across different behaviors, improving recommendation effectiveness. First, MB-SVD integrates the representation of users and items under different behaviors with learnable weight scores, which efficiently considers the influence of different behaviors. Then, MB-SVD generates augmented graph representation with global collaborative relations. Next, we simplify the contrastive learning framework by directly contrasting original representation with the enhanced representation using the InfoNCE loss. Through extensive experimentation, the remarkable performance of our proposed MB-SVD approach in multi-behavior recommendation endeavors across diverse real-world datasets is exhibited.
Paper Structure (19 sections, 14 equations, 3 figures, 2 tables)

This paper contains 19 sections, 14 equations, 3 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Statement
  4. METHODOLOGY
  5. Overview
  6. Behavior-guided Graph Representation
  7. Behavior-aware Aggregation
  8. Capturing Global Collaborative Dependencies with SVD
  9. Multi-behavior Contrastive Learning
  10. Loss Function and Joint Optimization
  11. Experimental Results and Analysis
  12. Dataset
  13. Baselines and Evaluation Metrics
  14. Experimental settings
  15. Main Results(RQ1)
  16. ...and 4 more sections

Figures (3)

  • Figure 1: Scenarios of multiple behavior recommendations in e-commerce. It illustrates the key elements of the scenarios, including users, items, and various types of behaviors. The red line represents purchase behavior, the green line indicates view behavior, and the yellow line represents add-to-cart behavior.
  • Figure 2: The architectural framework of MB-SVD. We illustrate the model with an instance where $K$ is equal to 3, indicating that there are three types of behaviors.
  • Figure 3: Hyperparameter effects of MB-SVD on three datasets.