Research map / Graph learning / Knowledge and temporal graphs
Composition-based Multi-Relational Graph Convolutional 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 |
|---|---|---|---|
| use CompGCN for link prediction | FB15k-237 · FB15k-237 | MRR: .35 -> .355 | Helped |
| use CompGCN for link prediction | WN18RR · WN18RR | MRR: .476 -> .479 | Helped |
| use CompGCN | MUTAG (Node) · MUTAG (Node) | accuracy: 80.9 -> 85.3 ± 1.2 | Helped |
| use CompGCN | AM · AM | accuracy: 90.2 ± 0.9 -> 90.6 ± 0.2 | Helped |
| use CompGCN | MUTAG (Graph) · MUTAG (Graph) | accuracy: 92.6 ± 4.2 -> 89.0 ± 11.1 | Mixed |
| use CompGCN | PTC · PTC | accuracy: 67.8 ± 13.2 -> 71.6 ± 12.0 | Helped |
| add circular-correlation composition instead of subtraction (ablation) | ConvE + CompGCN · FB15k-237 | MRR: 0.352 -> 0.355 | Helped |
| add circular-correlation composition instead of multiplication (ablation) | ConvE + CompGCN · FB15k-237 | MRR: 0.353 -> 0.355 | Helped |
| use 50 relation basis vectors rather than separate relation embeddings (ablation) | ConvE + CompGCN · FB15k-237 | MRR: 0.355 -> 0.350 | Hurt |
| use more relation basis vectors (ablation) | CompGCN · FB15k-237 | MRR | Helped |
| use CompGCN rather than W-GCN as the encoder | ConvE on FB15k-237 · FB15k-237 | MRR: 0.344 -> 0.355 | Helped |
Closest papers by meaning
- End-to-End Structure-Aware Convolutional Networks for Knowledge Base Completion (AAAI 2019 · similarity 0.90)
- Convolutional 2D Knowledge Graph Embeddings (AAAI 2018 · similarity 0.89)
- Factorizable Graph Convolutional Networks (NeurIPS 2020 · similarity 0.89)
- Distilling Knowledge From Graph Convolutional Networks (CVPR 2020 · similarity 0.89)
- Collaborative Graph Convolutional Networks: Unsupervised Learning Meets Semi-Supervised Learning (AAAI 2020 · similarity 0.89)
- Refining Latent Homophilic Structures over Heterophilic Graphs for Robust Graph Convolution Networks (AAAI 2024 · similarity 0.88)
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.