Disentangled Cascaded Graph Convolution Networks for Multi-Behavior Recommendation

TL;DR

Disen-CGCN addresses data sparsity and cold-start in multi-behavior recommendation by disentangling user and item factors within each behavior, cascading information across behaviors with a personalized feature transformation meta-network, and weighing factor-level preferences via attention. Built on LightGCN, it partitions embeddings into K factors per behavior, enforces independence with distance correlation, and uses a meta-network to generate behavior-specific transformation matrices, enabling personalized transfer. Empirically, it surpasses state-of-the-art baselines on Beibei and Tmall by substantial margins, with ablations confirming the contributions of disentanglement, attention, and personalization to the performance gains. The approach yields nuanced, user-centric recommendations and provides insights into how preferences shift across behaviors, offering practical impact for more effective multi-behavior recommender systems, with code released for reproducibility.

Abstract

Multi-behavioral recommender systems have emerged as a solution to address data sparsity and cold-start issues by incorporating auxiliary behaviors alongside target behaviors. However, existing models struggle to accurately capture varying user preferences across different behaviors and fail to account for diverse item preferences within behaviors. Various user preference factors (such as price or quality) entangled in the behavior may lead to sub-optimization problems. Furthermore, these models overlook the personalized nature of user behavioral preferences by employing uniform transformation networks for all users and items. To tackle these challenges, we propose the Disentangled Cascaded Graph Convolutional Network (Disen-CGCN), a novel multi-behavior recommendation model. Disen-CGCN employs disentangled representation techniques to effectively separate factors within user and item representations, ensuring their independence. In addition, it incorporates a multi-behavioral meta-network, enabling personalized feature transformation across user and item behaviors. Furthermore, an attention mechanism captures user preferences for different item factors within each behavior. By leveraging attention weights, we aggregate user and item embeddings separately for each behavior, computing preference scores that predict overall user preferences for items. Our evaluation on benchmark datasets demonstrates the superiority of Disen-CGCN over state-of-the-art models, showcasing an average performance improvement of 7.07% and 9.00% on respective datasets. These results highlight Disen-CGCN's ability to effectively leverage multi-behavioral data, leading to more accurate recommendations.

Figures (6)

• Figure 1: Illustration of user preference for different types of behaviors, where the boldness of each arrow indicates the degree of preference.
• Figure 2: Overview of our Disen-CGCN model.
• Figure 3: Visualize the different preferences of users in different behaviors.
• Figure 4: Impact of the number of factors (K).
• Figure 5: Impact of the number of GCN layers (L).