Research map / Graph learning / Knowledge and temporal graphs
End-to-End Structure-Aware Convolutional Networks for Knowledge Base Completion
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 |
|---|---|---|---|
| Replace ConvE with the translational Conv-TransE decoder | ConvE · FB15k-237 | Hits@10: 0.49 -> 0.51 | Helped |
| Add WGCN structure-aware encoding | Conv-TransE · FB15k-237 | Hits@10: 0.51 -> 0.54 | Helped |
| Add WGCN structure-aware encoding | Conv-TransE · WN18RR | MRR: 0.46 -> 0.47 | Helped |
| Add node attributes to SACN training | SACN · FB15k-237-Attr | Hits@10: 0.54 -> 0.55 | Helped |
| Add node attributes as entity triplets | ConvE · FB15k-237-Attr | Hits@10: 0.49 -> 0.46 | No clear effect |
| Increase kernel width from 1 to 3 (ablation) | Conv-TransE · FB15k-237 | MRR: 0.324 -> 0.331 | Helped |
| Increase kernel width from 3 to 5 (ablation) | Conv-TransE · FB15k-237 | MRR: 0.331 -> 0.329 | Hurt |
| Increase kernel width from 1 to 3 (ablation) | SACN · FB15k-237 | MRR: 0.345 -> 0.351 | Helped |
| Increase kernel width from 3 to 5 (ablation) | SACN · FB15k-237 | MRR: 0.351 -> 0.352 | Helped |
| Increase kernel width from 1 to 5 (ablation) | SACN+Attr · FB15k-237-Attr | MRR: 0.351 -> 0.360 | Helped |
| Use WGCN aggregation rather than no structure aggregation (ablation) | SACN versus Conv-TransE · FB15k-237, nodes with indegree [500,1000] | Hits@3: 0.767 -> 0.695 | Hurt |
| Use WGCN aggregation rather than no structure aggregation (ablation) | SACN versus Conv-TransE · FB15k-237, nodes with indegree [0,100] | Hits@3: 0.125 -> 0.134 | Helped |
Closest papers by meaning
- Convolutional 2D Knowledge Graph Embeddings (AAAI 2018 · similarity 0.92)
- Distilling Knowledge From Graph Convolutional Networks (CVPR 2020 · similarity 0.91)
- Composition-based Multi-Relational Graph Convolutional Networks (ICLR 2020 · similarity 0.90)
- Learning Graphs for Knowledge Transfer With Limited Labels (CVPR 2021 · similarity 0.89)
- Mixed Geometry Message and Trainable Convolutional Attention Network for Knowledge Graph Completion (AAAI 2024 · similarity 0.88)
- Towards Global-Topology Relation Graph for Inductive Knowledge Graph Completion (AAAI 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.