Graph Contrastive Learning for Multi-Behavior Recommender System
Year: 2024
Type: article
Abstract: Multi-behavior recommender system aims to model user preference representation based on multiple types of user-item interactions (e.g., viewing, adding to favorites, adding to the cart, and purchasing). However, existing works have two limitations in general: 1) Most of them only concern the sparse observed user-item interactions and ignore the huge amount of unobserved user-item interactions, Which is incapable of fully capturing user preference in recommender systems. 2) Previous works typically tend to only extract valuable information by distinguishing target and auxiliary behaviors to model user reference representation, and they fail to explore the fine-grained commonality between target and auxiliary behaviors. To tackle these limitations, we propose a new model named graph contrastive learning with multi-behavior (GCMB) for a multi-behavior recommender system. Specifically, we devise a graph contrastive learning paradigm guided by Randomized Singular Value Decomposition (RSVD) instead of the traditional three-view paradigm to inject the global collaborative relationship (observed and unobserved user-item interactions) into the multi-behavior recommender model. After that, we can utilize the global collaborative relationship to more effectively capture user preference preference in the original view. Furthermore, we consider the multi-level commonality between target and auxiliary behaviors to capture the fine-grained commonality and then model high quality of user preference representation. Extensive experiments on two real-world datasets demonstrate that our method consistently outperforms various state-of-the-art recommender methods.
Language: English
Subfield: Information Systems
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: G1121966298, G3492796586, G5457408167
OpenAlex ID: W4405022935