Research map / Graph learning / GNN architectures
Relation Extraction with Convolutional Network over Learnable Syntax-Transport Graph
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 node-importance weights AD (ablation) | LST-AGCN · SemEval-2010 Task 8 | macro-F1: 84.7 -> 86.0 | Helped |
| add node-importance weights AD (ablation) | LST-AGCN · TACRED | micro-F1: 66.7 -> 68.8 | Helped |
| learn edge transport weights AT instead of binary adjacency (ablation) | LST-AGCN · SemEval-2010 Task 8 | macro-F1: 84.9 -> 86.0 | Helped |
| learn edge transport weights AT instead of binary adjacency (ablation) | LST-AGCN · TACRED | micro-F1: 67.8 -> 68.8 | Helped |
| add attention over GCN layers (ablation) | LST-AGCN · SemEval-2010 Task 8 | macro-F1: 85.5 -> 86.0 | Helped |
| add attention over GCN layers (ablation) | LST-AGCN · TACRED | micro-F1: 68.6 -> 68.8 | Helped |
| add GCN over BiLSTM embeddings (ablation) | LST-AGCN · TACRED | micro-F1: 57.1 -> 68.8 | Helped |
| add GCN over BiLSTM embeddings (ablation) | LST-AGCN · SemEval-2010 Task 8 | macro-F1: 84.6 -> 86.0 | Helped |
| learn transport matrix instead of binary adjacency (LST-GCN K=3 vs ST-GCN) | LST-AGCN · SemEval-2010 Task 8 | macro-F1: 85.0 -> 85.5 | Helped |
| learn transport matrix instead of binary adjacency (LST-GCN K=2 vs ST-GCN) | LST-AGCN · TACRED | micro-F1: 67.2 -> 68.6 | Helped |
| increase GCN layers K from 2 to 4 without attention | LST-AGCN · TACRED | micro-F1: 68.6 -> 65.9 | Hurt |
| LST-AGCN vs C-AGGCN | LST-AGCN · SemEval-2010 Task 8 | macro-F1: 85.7 -> 86.0 | Helped |
Closest papers by meaning
- GATE: Graph Attention Transformer Encoder for Cross-lingual Relation and Event Extraction (AAAI 2021 · similarity 0.88)
- Dependency Structure-Enhanced Graph Attention Networks for Event Detection (AAAI 2024 · similarity 0.87)
- Composition-based Multi-Relational Graph Convolutional Networks (ICLR 2020 · similarity 0.86)
- GDPNet: Refining Latent Multi-View Graph for Relation Extraction (AAAI 2021 · similarity 0.86)
- Factorizable Graph Convolutional Networks (NeurIPS 2020 · similarity 0.85)
- End-to-End Structure-Aware Convolutional Networks for Knowledge Base Completion (AAAI 2019 · 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.