Research map / Graph learning / Knowledge and temporal graphs
Neural Compositional Rule Learning for Knowledge Graph Reasoning
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 |
|---|---|---|---|
| learn compositional rules with NCRL rather than the strongest reported rule-learning baseline | KG completion · WN18RR | MRR: 0.47 -> 0.67 | Helped |
| learn compositional rules with NCRL rather than the strongest reported rule-learning baseline | KG completion · UMLS | MRR: 0.75 -> 0.78 | Helped |
| use NCRL instead of the strongest reported baseline | KG completion · Kinship | MRR: 0.65 -> 0.64 | Hurt |
| use NCRL instead of the strongest reported baseline | KG completion · FB15K-237 | MRR: 0.32 -> 0.30 | Hurt |
| add learned hierarchical structure selection instead of using a random deduction order (ablation) | NCRL · Family | MRR: 0.84 -> 0.92 | Helped |
| add learned hierarchical structure selection instead of using a random deduction order (ablation) | NCRL · Kinship | MRR: 0.34 -> 0.65 | Helped |
| add learned hierarchical structure selection instead of using a random deduction order (ablation) | NCRL · UMLS | MRR: 0.55 -> 0.78 | Helped |
| set the sliding-window size to 2 rather than 3 (ablation) | NCRL · Family | MRR: 0.90 -> 0.92 | Helped |
| set the sliding-window size to 2 rather than 3 (ablation) | NCRL · Kinship | MRR: 0.60 -> 0.65 | Helped |
| use an RNN rather than an MLP as the sliding-window encoder (ablation) | NCRL · Kinship | MRR: 0.56 -> 0.65 | Helped |
| use an RNN rather than a Transformer as the sliding-window encoder (ablation) | NCRL · Family | MRR: 0.84 -> 0.92 | Helped |
| train NCRL rather than Neural-LP | KG completion training · WN18RR | training time: 1,308 -> 77 | Helped |
Closest papers by meaning
- Rule-Guided Graph Neural Networks for Explainable Knowledge Graph Reasoning (AAAI 2025 · similarity 0.88)
- Understanding Expressivity of GNN in Rule Learning (ICLR 2024 · similarity 0.87)
- Rethinking Complex Queries on Knowledge Graphs with Neural Link Predictors (ICLR 2024 · similarity 0.87)
- Improving Complex Reasoning over Knowledge Graph with Logic-Aware Curriculum Tuning (AAAI 2025 · similarity 0.86)
- Topology of Reasoning: Understanding Large Reasoning Models through Reasoning Graph Properties (NeurIPS 2025 · similarity 0.86)
- Boule or Baguette? A Study on Task Topology, Length Generalization, and the Benefit of Reasoning Trace (NeurIPS 2026 · similarity 0.86)
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.