Research map / Graph learning / GNN architectures
SDGNN: Learning Node Representation for Signed Directed Networks
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 |
|---|---|---|---|
| Model signs, directions, and triangles jointly (ablation) | SDGNN · Bitcoin-Alpha | Macro-F1: 0.6738 -> 0.7585 | Helped |
| Add direction loss to sign reconstruction (ablation) | SDGNN · Bitcoin-Alpha | Macro-F1: 0.6738 -> 0.7414 | Helped |
| Add triangle loss to sign reconstruction (ablation) | SDGNN · Bitcoin-Alpha | Macro-F1: 0.6738 -> 0.7210 | Helped |
| Use attention instead of mean aggregation (ablation) | 2-layer SDGNN · Bitcoin-Alpha | AUC: 0.9041 -> 0.9154 | Helped |
| Use two aggregation layers instead of one (ablation) | GAT-aggregated SDGNN · Bitcoin-Alpha | AUC: 0.9095 -> 0.9154 | Helped |
| Use SDGNN instead of BESIDE | link sign prediction · Bitcoin-OTC | Macro-F1: 0.7843 -> 0.8017 | Helped |
| Use SDGNN instead of BESIDE | link sign prediction · Bitcoin-OTC | AUC: 0.9152 -> 0.9124 | Hurt |
| Use SDGNN instead of BESIDE | link sign prediction · Wikirfa | Macro-F1: 0.7803 -> 0.7849 | Helped |
| Use SDGNN instead of BESIDE | link sign prediction · Wikirfa | AUC: 0.8981 -> 0.8898 | Hurt |
| Use SDGNN instead of BESIDE | link sign prediction · Slashdot | Macro-F1: 0.7892 -> 0.7892 | No clear effect |
| Use SDGNN instead of BESIDE | link sign prediction · Slashdot | AUC: 0.9017 -> 0.8977 | Hurt |
| Use SDGNN instead of FeExtra | link sign prediction · Epinions | AUC: 0.9444 -> 0.9411 | Hurt |
Closest papers by meaning
- Self-Explainable Graph Transformer for Link Sign Prediction (AAAI 2025 · similarity 0.89)
- Structure Balance and Gradient Matching-Based Signed Graph Condensation (AAAI 2025 · similarity 0.89)
- Sign is Not a Remedy: Multiset-to-Multiset Message Passing for Learning on Heterophilic Graphs (ICML 2024 · similarity 0.88)
- Bridging Theory and Practice in Link Representation with Graph Neural Networks (NeurIPS 2025 · similarity 0.88)
- DropEdge not Foolproof: Effective Augmentation Method for Signed Graph Neural Networks (NeurIPS 2024 · similarity 0.88)
- A Signed Graph Approach to Understanding and Mitigating Oversmoothing (NeurIPS 2025 · similarity 0.87)
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.