Research map / Graph learning / Knowledge and temporal graphs
Coordinated Reasoning for Cross-Lingual Knowledge Graph Alignment
Kun Xu, Linfeng Song, Yansong Feng, Yan Song, Dong Yu
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 |
|---|---|---|---|
| Apply Easy-to-Hard decoding and joint entity alignment | GM · DBP15K ZH-EN | Hits@1: 67.93 -> 73.58 | Improved |
| Apply Easy-to-Hard decoding and joint entity alignment | GM · DBP15K JA-EN | Hits@1: 73.97 -> 79.15 | Improved |
| Apply Easy-to-Hard decoding and joint entity alignment | GM · DBP15K FR-EN | Hits@1: 89.38 -> 92.43 | Improved |
| Add joint entity alignment | GCN, BootEA, and RDGCN · DBP15K ZH-EN, JA-EN, and FR-EN | Hits@1: GCN: 41.25 -> 43.43, 39.9 -> 45.00, 37.29 -> 39.78; BootEA: 62.94 -> 64.56, 62.23 -> 64.17, 65.30 -> 69.31; RDGCN: 70.75 | Improved |
| Add Easy-to-Hard decoding | GCN and BootEA · DBP15K ZH-EN, JA-EN, and FR-EN | Hits@1: GCN: 41.25 -> 44.37, 39.9 -> 41.72, 37.29 -> 39.09; BootEA: 62.94 -> 65.27, 62.23 -> 65.36, 65.30 -> 68.92 | Improved |
| Add Easy-to-Hard decoding | RDGCN · DBP15K ZH-EN, JA-EN, and FR-EN | Hits@1: 70.75 -> 71.15, 76.74 -> 77.07, 88.64 -> 91.01 | Mixed |
| Use Easy-to-Hard decoding without joint entity alignment (ablation) | GM · DBP15K ZH-EN, JA-EN, and FR-EN | Hits@1: 67.93 -> 70.31, 73.97 -> 77.92, 89.38 -> 90.49 | Improved |
| Use joint entity alignment without Easy-to-Hard decoding (ablation) | GM · DBP15K ZH-EN, JA-EN, and FR-EN | Hits@1: 67.93 -> 72.05, 73.97 -> 78.73, 89.38 -> 91.08 | Improved |
| Lower the Easy-to-Hard threshold from α=0.75 to α=0.65 (ablation) | GM Easy-to-Hard decoding · DBP15K development | Hits@1: 72.09 -> 67.12 | Hurt |
| Raise the Easy-to-Hard threshold from α=0.75 to α=0.95 (ablation) | GM Easy-to-Hard decoding · DBP15K ZH-EN, JA-EN, and FR-EN development | Hits@1: 72.09 -> 69.05; 76.62 -> 74.33; 91.18 -> 88.25 | Hurt |
| Reduce the joint-alignment candidate cutoff from τ=0.15 to τ=0.10 (ablation) | Joint entity alignment · DBP15K FR-EN | Hits@1: 90.90 -> 91.02 | Improved |
| Reduce the joint-alignment candidate cutoff from τ=0.10 to τ=0.05 (ablation) | Joint entity alignment · DBP15K FR-EN | runtime: 24m34s; roughly 32 hours | Hurt |
Closest papers by meaning
- HLMEA: Unsupervised Entity Alignment Based on Hybrid Language Models (AAAI 2025 · similarity 0.91)
- Entity Alignment with Noisy Annotations from Large Language Models (NeurIPS 2024 · similarity 0.90)
- NeuSymEA: Neuro-symbolic Entity Alignment via Variational Inference (NeurIPS 2025 · similarity 0.88)
- A Systematic Exploration of Knowledge Graph Alignment with Large Language Models in Retrieval Augmented Generation (AAAI 2025 · similarity 0.88)
- Lambda: Learning Matchable Prior For Entity Alignment with Unlabeled Dangling Cases (NeurIPS 2024 · similarity 0.87)
- Soft Token Alignment for Cross-Lingual Reasoning (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.
