Research map / Label-efficient and robust learning / Zero-/few-shot and meta-learning
Inductive Relation Prediction by BERT
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 |
|---|---|---|---|
| Add reasoning paths to BERT input | KG-BERT / BERTRL · FB15k-237 inductive, full | Hits@1: 0.341 -> 0.541 | Helped |
| Add reasoning paths to BERT input | KG-BERT / BERTRL · NELL-995 inductive, full | Hits@1: 0.244 -> 0.715 | Helped |
| Add BERT-based reasoning paths | Knowledge graph completion · WN18RR transductive | Hits@1: 0.600 -> 0.655 | Helped |
| Use edge-list subgraph linearization instead of individual paths (ablation) | BERTRL · FB15k-237 inductive | Hits@1: 0.441 -> 0.361; 0.493 -> 0.398; 0.541 -> 0.463 | Hurt |
| Combine reasoning paths into one input instead of separate instances (ablation) | BERTRL · FB15k-237 inductive | Hits@1: 0.441 -> 0.351; 0.493 -> 0.461; 0.541 -> 0.505 | Hurt |
| Use 5 sampled individual paths (ablation) | BERTRL · FB15k-237 inductive | Hits@1: 0.441 -> 0.466; 0.493 -> 0.490; 0.541 -> 0.532 | Mixed |
| Use 10 sampled individual paths (ablation) | BERTRL · FB15k-237 inductive | Hits@1: 0.441 -> 0.449; 0.493 -> 0.505; 0.541 -> 0.500 | Mixed |
| Use pre-trained BERT for unseen-relation generalization | FB15k-237 · 50 unseen relations | Hits@1: 0.266 -> 0.485 | Helped |
Closest papers by meaning
- Inductive Relation Prediction Using Analogy Subgraph Embeddings (ICLR 2022 · similarity 0.90)
- Inducing Relational Knowledge from BERT (AAAI 2020 · similarity 0.89)
- Inductive Relation Prediction by Subgraph Reasoning (ICML 2020 · similarity 0.89)
- Towards Global-Topology Relation Graph for Inductive Knowledge Graph Completion (AAAI 2025 · similarity 0.86)
- Anchoring Path for Inductive Relation Prediction in Knowledge Graphs (AAAI 2024 · similarity 0.85)
- Towards Foundation Models for Knowledge Graph Reasoning (ICLR 2024 · similarity 0.85)
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.