Research map / Graph learning / Graph algorithms and clustering
Ensembling Graph Predictions for AMR Parsing
Thanh Lam Hoang, Gabriele Picco, Yufang Hou, Young-Suk Lee, Lam M. Nguyen, Dzung T. Phan, Vanessa López, Ramon Fernandez Astudillo
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 |
|---|---|---|---|
| ensemble four SPRING checkpoints with Graphene | SPRING · AMR 2.0 test | Smatch: 84.22 -> 84.78 | Improved |
| add APT, T5, Cai&Lam models to the ensemble (Graphene Support vs 4S) | Graphene 4S · AMR 2.0 test | Smatch: 84.78 -> 85.85 | Improved |
| Graphene ensemble of all models vs best single model | SPRING · AMR 3.0 test | Smatch: 83.25 -> 84.41 | Improved |
| Graphene ensemble on out-of-distribution BIO | SPRING · BIO | Smatch: 60.52 -> 62.29 | Improved |
| select candidate by Smatch instead of support (Graphene Smatch) | Graphene Support · AMR 2.0 test | Smatch: 85.85 -> 86.26 | Improved |
| create pivot-modified graphs versus Barzdins selection only | Barzdins · New3 | Smatch: 75.87 -> 76.32 | Improved |
| use support as selection criterion instead of Smatch | Graphene Smatch · BIO | Smatch: 62.80 -> 62.29 | Mixed |
| consider every prediction as pivot instead of only SPRING pivot (ablation) | Graphene SPRING pivot · AMR 2.0 | Smatch: 84.70 -> 85.85 | Improved |
| use SPRING as sole pivot vs plain SPRING prediction (ablation) | SPRING · BIO | Smatch: 60.52 -> 61.56 | Improved |
| support threshold theta 0.5 instead of 0.1 (ablation) | Graphene · AMR 2.0 dev | Smatch: 81.64 -> 85.49 | Improved |
| support threshold theta 0.9 instead of 0.5 (ablation) | Graphene · AMR 2.0 dev | Smatch: 85.49 -> 83.68 | Hurt |
| compare Graphene to uniform sampling baseline (ablation) | uniform Sampling · BIO | Smatch: 56.00 -> 62.29 | Improved |
Closest papers by meaning
- Ensemble Distillation for Unsupervised Constituency Parsing (ICLR 2024 · similarity 0.84)
- Ensembling Language Models with Sequential Monte Carlo (NeurIPS 2026 · similarity 0.84)
- Determine-Then-Ensemble: Necessity of Top-k Union for Large Language Model Ensembling (ICLR 2025 · similarity 0.84)
- Rethinking LLM Ensembling from the Perspective of Mixture Models (ICML 2026 · similarity 0.83)
- Graph Parsing Networks (ICLR 2024 · similarity 0.83)
- Talk like a Graph: Encoding Graphs for Large Language Models (ICLR 2024 · similarity 0.82)
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.
