Abstract:A well-developed recommendation system can not only leverage multi-typed interactions (such as page view, add-to-cart, and purchase) to better identify user preferences, but also demonstrate high performance, low complexity, and strong interpretability. However, many existing solutions for multi-behavior recommendation fall short of intuitive modeling of real-world scenarios, leading to overly complex models with massive parameters and cumbersome components. In particular, they share two critical limitations: (1) Some pioneering models are built upon the strict assumption of cascade effects across behaviors, which contradicts multifarious behavior paths in practical applications. (2) Existing approaches fail to explicitly capture the unique idiosyncrasies of users and even neglect the inherent nature of items involved in the multi-behavior interactions. To this end, we propose a novel Directed Acyclic Graph Convolutional Network (DA-GCN) for the multi-behavior recommendation task. Specifically, we pinpoint the partial order relations within the monotonic behavior chain and extend it to personalized directed acyclic behavior graphs to exploit behavior dependencies. Then, a GCN-based directed edge encoder is employed to distill rich collaborative signals embodied by each directed edge. In light of the information flows over the directed acyclic structure, we propose an attentive aggregation module to gather messages from all potential antecedent behaviors, representing distinct perspectives to understand the terminated behavior. Thus, we obtain comprehensive representations for the follow-up behavior through learnable distributions over its preceding behaviors, explicitly reflecting personalized interactive patterns of users and underlying properties of items simultaneously. Finally, we design a customized multi-task learning objective for flexible joint optimization. Extensive experiments on public benchmarking datasets fully demonstrate the superiority of DA-GCN with significant performance improvement and computational efficiency over a wide range of state-of-the-art methods. Our code is available at https://github.com/xizhu1022/DA-GCN.

MBDL: Exploring Dynamic Dependency among Various Types of Behaviors for Recommendation

Multi-Behavior Dynamic Contrastive Learning for Recommendation

Coarse-to-Fine Knowledge-Enhanced Multi-Interest Learning Framework for Multi-Behavior Recommendation

Contrastive Meta Learning with Behavior Multiplicity for Recommendation

Multi-behavior recommendation based on intent learning

Dual Graph Attention Networks for Multi-Behavior Recommendation

BPMCF: Behavior Preference Mapping Collaborative Filtering for Multi-behavior Recommendation.

A Multi-Behavior Recommendation Method Exploring the Preference Differences among Various Behaviors.

MBN: Towards Multi-Behavior Sequence Modeling for Next Basket Recommendation

Graph Meta Network for Multi-Behavior Recommendation

Improving Implicit Feedback-Based Recommendation through Multi-Behavior Alignment

Dual Contrastive Learning for Multi-Behavior Recommendation

Dual-view co-contrastive learning for multi-behavior recommendation

Knowledge Enhancement for Contrastive Multi-Behavior Recommendation

Multi-view multi-behavior interest learning network and contrastive learning for multi-behavior recommendation

Multi-Behavior Graph Neural Networks for Recommender System

Multi-behavioral recommendation model based on dual neural networks and contrast learning

Multiple Behaviors Recommendation with Graph Learning

Multi-Behavior Recommendation with Personalized Directed Acyclic Behavior Graphs

Learning to Recommend with Multiple Cascading Behaviors

Multi-Behavior Contrastive Learning with Graph Neural Networks for Recommendation