Abstract:Heterogeneous information network (HIN) has been applied in a wide variety of graph analysis tasks. At present, it is a trend of heterogeneous graph neural networks (HGNNs) to cast the meta-paths aside, since it solves the problem of structural information loss caused by artificially designed meta-paths. However, existing meta-path-free HGNNs fail to take into account that most node types in many HINs have no attributes, and they cannot make full use of sparse node attributes when applied to HINs with missing attributes. Furthermore, their computation of attention coefficients explores the correlations of node attributes while almost ignoring structural ones, which may limit the expression ability of the model and cause overfitting in model training. To alleviate these issues, we propose an HGNN with attribute enhancement and structure-aware attention (HGNN-AESA). First, we design an attribute enhancement module (AEM) to connect more useful attributed nodes to the target nodes. Specifically, AEM introduces a random walk with restart (RWR) strategy to obtain structural relevance scores of each node within its specific subgraph. The structural relevance scores are used to capture potentially influential attributed nodes in high-order neighborhood for each target node. Second, we propose heterogeneous structure-aware attention layers (HSALs) to learn node representations. HSALs follow a hierarchical attention framework, including node-level and type-level attention. The node-level attention aggregates feature (attribute) embeddings of same-type neighbors, and the relevant attention coefficients depend on the combination of node attributes and heterogeneous structural interventions. The type-level attention fuses all type-specific vector representations and generates the ultimate node embedding. Finally, extensive experiments on three different real-world HIN datasets demonstrate that our model outperforms state-of-the-art methods.

Adversarial Heterogeneous Network Embedding with Metapath Attention Mechanism

Heterogeneous Information Network Embedding with Meta-path Based Graph Attention Networks.

Adversarial Learning on Heterogeneous Information Networks

Attention Based Meta Path Fusion for Heterogeneous Information Network Embedding

Heterogeneous Information Network Embedding with Adversarial Disentangler

Generative Adversarial Network and Meta-path Based Heterogeneous Network Representation Learning

Heterogeneous Information Network Embedding with Convolutional Graph Attention Networks.

Meta-path-based heterogeneous graph neural networks in academic network

Heterogeneous Graph Fully-Connected Attention Network

Heterogeneous Graph Embedding by Aggregating Meta-Path and Meta-Structure Through Attention Mechanism

Metapath Enhanced Graph Attention Encoder For Hins Representation Learning

Adaptive Neighbor Graph Aggregated Graph Attention Network for Heterogeneous Graph Embedding

Gahne: Graph-Aggregated Heterogeneous Network Embedding

A Heterogeneous Graph Neural Network with Attribute Enhancement and Structure-Aware Attention

HAHE: Hierarchical Attentive Heterogeneous Information Network Embedding.

Meta-neighbor Aggregated Graph Attention Network for Heterogeneous Graph Representation

Adversarial Mutual Information Learning for Network Embedding

Meta-HGT: Metapath-aware HyperGraph Transformer for heterogeneous information network embedding

A 2 CMHNE: Attention-Aware Collaborative Multimodal Heterogeneous Network Embedding

GANE: A Generative Adversarial Network Embedding

Attributed Heterogeneous Graph Embedding with Meta-graph Attention