Research map / Graph learning / Knowledge and temporal graphs
Simple and Efficient Heterogeneous Graph Neural Network
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Remove neighbor attention (ablation) | HAN on DBLP · DBLP | macro-F1: 92.59 -> 92.75 | No clear effect |
| Remove semantic attention (ablation) | HAN on DBLP · DBLP | macro-F1: 92.59 -> 92.19 | Hurt |
| Use one layer with long metapaths rather than four layers with short metapaths (ablation) | HAN · DBLP | macro-F1: 89.55 -> 93.60 | Helped |
| Use one layer with long metapaths rather than four layers with short metapaths (ablation) | HAN · ACM | macro-F1: 88.62 -> 92.82 | Helped |
| Replace mean neighbor aggregation with GAT attention (ablation) | SeHGNN · DBLP | macro-F1: 95.06±0.17 -> 93.61±0.51 | Hurt |
| Use two layers with shorter metapaths per layer (ablation) | SeHGNN · DBLP | macro-F1: 95.06±0.17 -> 94.66±0.27 | Hurt |
| Add label propagation inputs (ablation) | SeHGNN · DBLP | macro-F1: 94.86±0.14 -> 95.06±0.17 | Helped |
| Use transformer semantic fusion instead of weighted-sum fusion (ablation) | SeHGNN · DBLP | macro-F1: 94.52±0.05 -> 95.06±0.17 | Helped |
| Add extra embeddings (ablation) | SeHGNN on ogbn-mag · ogbn-mag test | accuracy: 53.99±0.18 -> 54.78±0.17 | Helped |
| Use multi-stage training (ablation) | SeHGNN on ogbn-mag · ogbn-mag test | accuracy: 53.99±0.18 -> 56.71±0.14 | Helped |
| Add extra embeddings and multi-stage training (ablation) | SeHGNN on ogbn-mag · ogbn-mag test | accuracy: 53.99±0.18 -> 57.19±0.12 | Helped |
Closest papers by meaning
- Heterogeneous Graph Neural Network on Semantic Tree (AAAI 2025 · similarity 0.92)
- Bootstrapping Heterogeneous Graph Representation Learning via Large Language Models: A Generalized Approach (AAAI 2025 · similarity 0.91)
- An Efficient Subgraph-Inferring Framework for Large-Scale Heterogeneous Graphs (AAAI 2024 · similarity 0.91)
- HGOT: Self-supervised Heterogeneous Graph Neural Network with Optimal Transport (ICML 2025 · similarity 0.90)
- Contrastive Auxiliary Learning with Structure Transformation for Heterogeneous Graphs (AAAI 2025 · similarity 0.90)
- HeterSEED: Semantics–Structure Decoupling for Heterogeneous Graph Learning under Heterophily (NeurIPS 2026 · similarity 0.90)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.