Heterogeneous Graph Contrastive Learning for Recommendation

Abstract

Graph Neural Networks (GNNs) have become powerful tools in modeling graph-structured data in recommender systems. However, real-life recommendation scenarios usually involve heterogeneous relationships (e.g., social-aware user influence, knowledge-aware item dependency) which contains fruitful information to enhance the user preference learning. In this paper, we study the problem of heterogeneous graph-enhanced relational learning for recommendation. Recently, contrastive self-supervised learning has become successful in recommendation. In light of this, we propose a Heterogeneous Graph Contrastive Learning (HGCL), which is able to incorporate heterogeneous relational semantics into the user-item interaction modeling with contrastive learning-enhanced knowledge transfer across different views. However, the influence of heterogeneous side information on interactions may vary by users and items. To move this idea forward, we enhance our heterogeneous graph contrastive learning with meta networks to allow the personalized knowledge transformer with adaptive contrastive augmentation. The experimental results on three real-world datasets demonstrate the superiority of HGCL over state-of-the-art recommendation methods. Through ablation study, key components in HGCL method are validated to benefit the recommendation performance improvement. The source code of the model implementation is available at the link https://github.com/HKUDS/HGCL.

Multi-Label Zero-Shot Product Attribute-Value Extraction

“…To handle different types of nodes and edges, heterogeneous hypergraphs are learned by attention mechanisms [13,30,33,40], wavelets [59], and variational auto-encoder [14,39]. Though, all of these works are widely applied for social networks [34,64], academic citations [65,67,75], biological networks [25,44] or product recommendation in e-commerce [4,8,38,71], heterogeneous hypergraphs are never applied to attribute value extraction in e-commerce. Different from the above hypergraphs that build hyperedges by close neighbors or meta-paths, we construct e-commerce related hyperedges by using user behavior and product inventory data to capture higher-order relations among categories, products, and aspects, to recognize unseen attribute values for new products.…”

Graph-less Collaborative Filtering

“…Self-Supervised Learning (SSL) for Recommendation. To tackle the challenge of noise and sparsity in recommendation systems, recent research has explored various types of SSL techniques for data augmentation [6,30,35,41]. For instance, some studies, such as SGL [32], introduce random perturbation to generate additional views for CL.…”

Multi-Relational Contrastive Learning for Recommendation

“…Contrastive Representation Learning. Contrastive learning has emerged as a popular technique for data augmentation with auxiliary self-supervised signals [6,16,24,44]. In the domain of image analysis, numerous contrastive learning methods have been proposed for modeling image data with different augmentation techniques [1,33,38], such as cropping, horizontal translations, and rotations.…”