Research map / Graph learning / GNN architectures
Line Graph Neural Networks for Link Prediction
Lei Cai, Jundong Li, Jie Wang, Shuiwang Ji
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| learn link features in the line graph instead of enclosing-subgraph classification with pooling (LGLP vs SEAL) | link prediction GNN · C.ele (80% training links) | AUC: 87.44 -> 90.16 | Improved |
| learn link features in the line graph (LGLP vs SEAL) | link prediction GNN · BUP (50% training links) | AUC: 85.10 -> 88.57 | Improved |
| learn link features in the line graph (LGLP vs SEAL) | link prediction GNN · EML (80% training links) | AUC: 92.01 -> 92.03 | No clear effect |
| add node attributes as input | SEAL · Cora (50% training links) | AUC: 79.95 -> 75.33 | Hurt |
| add node attributes as input | LGLP · Cora (50% training links) | AUC: 79.96 -> 81.45 | Improved |
| add node attributes with LGLP combination function instead of SEAL concatenation | link prediction GNN · Cora with attributes (50% training links) | AUC: 75.33 -> 81.45 | Improved |
| reduce training links from 80% to 50% | LGLP · USAir | AUC: 97.44 -> 95.18 | Hurt |
| reduce training links from 80% to 50% | SEAL · USAir | AUC: 95.21 -> 93.23 | Hurt |
| use line graph with graph convolution only, no pooling, instead of SEAL pooling | LGLP · training convergence | epochs: 50 -> 10 to 15 | Improved |
Closest papers by meaning
- A Multi-Scale Approach for Graph Link Prediction (AAAI 2020 · similarity 0.91)
- Graph Neural Networks for Link Prediction with Subgraph Sketching (ICLR 2023 · similarity 0.90)
- Neural Link Prediction with Walk Pooling (ICLR 2022 · similarity 0.90)
- A Topological Perspective on Demystifying GNN-Based Link Prediction Performance (ICLR 2024 · similarity 0.89)
- GNNs Meet Sequence Models Along the Shortest-Path: an Expressive Method for Link Prediction (NeurIPS 2026 · similarity 0.88)
- Implicit degree bias in the link prediction task (ICML 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.
