Jing Yan

Jing Yan

Auto-generated profile, based on publicly available information

Jing Yan

Department of Computer Science and Technology, Tangshan Normal University, Tangshan, China

Research Focus & Metrics

Academic Career Length

2Years (2025-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
10 April 2025 in IEEE Access
Verified
Journal Article
Open access
YSYifan Sun
Yifan Sun
JYJing Yan
Jing Yan
LLLilei Lu
Lilei Lu
HZHongbo Zhang
Hongbo Zhang
YSYanhong Shang
Yanhong Shang
Compared to traditional heterogeneous graph (HG) data, Bi-typed Multi-relational Heterogeneous Graph (BMHG) not only have various edge relationships between different types of nodes but also connections among the same type of nodes, which increases modeling complexity. However, existing Heterogeneous Graph Neural Networks (HGNN) primarily focus on HGs with single relationships and are ineffective for BMHGs. These methods often rely on meta-paths to capture node features, which leads to the loss of intermediate node information. To address this, this paper proposes a Semantic Fusion-based Bi-typed Multi-relational Heterogeneous Graph model (SF-BMHG), aimed at solving the challenges of multiple edge relationships and complex semantic information processing among nodes in BMHGs. SF-BMHG first generates new features with intermediate nodes through a feature fusion strategy to avoid the loss of intermediate node information. It then captures semantic information in BMHGs using the new node features and meta-relation paths, integrating features from different semantic paths with a semantic-level attention mechanism. Additionally, it employs relational convolutions to capture relationship features within different types and fuses different relationship features through a relational-level attention mechanism. Ultimately, SF-BMHG combines intra-type relationship features and inter-type semantic features to produce rich feature embeddings. Experimental results on real-world constructed datasets demonstrate that, compared to existing baseline methods, SF-BMHG achieves superior performance across various tasks.