Heterogeneous Hypergraph Embedding for Recommendation Systems

TL;DR

Knowledge graphs boost recommender systems but often miss higher-order interactions and struggle with heterogeneous inputs. KHGRec introduces a Collaborative Knowledge Heterogeneous Hypergraph (CKHG) and dual encoders (Local Self-aware Hypergraph Encoder and Global Relational-aware Hypergraph Encoder) with attention-based feature fusion and cross-view contrastive learning to capture group-wise interactions and complex relational dependencies. Empirical results on four real-world datasets show consistent improvements over baselines, with an average $5.18\%$ relative gain in ranking metrics and demonstrated robustness to noise, cold-start, and data sparsity, while providing explainable recommendations via attention paths. The approach offers practical impact for scalable, explainable KG-enhanced recommendations and opens avenues for extensions with streaming data, trust-aware settings, and integration with pre-trained models or LLM-based augmentations.

Abstract

Recent advancements in recommender systems have focused on integrating knowledge graphs (KGs) to leverage their auxiliary information. The core idea of KG-enhanced recommenders is to incorporate rich semantic information for more accurate recommendations. However, two main challenges persist: i) Neglecting complex higher-order interactions in the KG-based user-item network, potentially leading to sub-optimal recommendations, and ii) Dealing with the heterogeneous modalities of input sources, such as user-item bipartite graphs and KGs, which may introduce noise and inaccuracies. To address these issues, we present a novel Knowledge-enhanced Heterogeneous Hypergraph Recommender System (KHGRec). KHGRec captures group-wise characteristics of both the interaction network and the KG, modeling complex connections in the KG. Using a collaborative knowledge heterogeneous hypergraph (CKHG), it employs two hypergraph encoders to model group-wise interdependencies and ensure explainability. Additionally, it fuses signals from the input graphs with cross-view self-supervised learning and attention mechanisms. Extensive experiments on four real-world datasets show our model's superiority over various state-of-the-art baselines, with an average 5.18\% relative improvement. Additional tests on noise resilience, missing data, and cold-start problems demonstrate the robustness of our KHGRec framework. Our model and evaluation datasets are publicly available at \url{https://github.com/viethungvu1998/KHGRec}.

Figures (11)

• Figure 1: An illustration of knowledge-assisted recommender systems.
• Figure 2: Overview of our framework for KGHRec, which consists of three main components: Heterogeneous hypergraph construction, Heterogeneous hypergraph representation learning, and Feature fusion module.
• Figure 3: Snapshots generation for CKHG.
• Figure 4: Architecture of Hypergraph Transformer Self-Attention Networks Convolution Layer.
• Figure 5: Architecture of Relational-aware Hypergraph Attention Convolution Layer.

Theorems & Definitions (3)

• Definition 1: Collaborative knowledge graph
• Definition 2: Heterogeneous Hypergraph
• Definition 3: CKHG