Research map / Graph learning / Knowledge and temporal graphs
GATE: Graph Attention Transformer Encoder for Cross-lingual Relation and Event Extraction
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 |
|---|---|---|---|
| use syntactic-distance attention instead of standard Transformer self-attention | GATE encoder · ACE05 single-source transfer | F-score (%): average improvement of 4.7% and 1.3% | Helped |
| use syntactic-distance attention instead of sequential relative-position representations | GATE encoder · ACE05 single-source transfer | F-score (%): average improvement of 1.3% and 1.9% | Helped |
| use GATE rather than Transformer_RPR | ACE05 En→Zh single-source relation extraction · ACE05 En→Zh | F-score (%): 58.0 -> 55.1 | Hurt |
| use GATE rather than Transformer_RPR | ACE05 {En,Ar}→Zh multi-source event argument role labeling · ACE05 {En,Ar}→Zh | F-score (%): 68.4 -> 65.3 | Hurt |
| use GATE rather than Transformer_RPR | ACE05 {En,Ar}→Zh multi-source relation extraction · ACE05 {En,Ar}→Zh | F-score (%): 62.3 -> 57.9 | Hurt |
| replace M-BERT word features with XLM-R (ablation) | GATE · ACE05 En→Ar event argument role labeling | F-score (%): 54.8 -> 61.7 | Helped |
| replace M-BERT word features with multilingual word embeddings (ablation) | GATE · ACE05 En→Zh event argument role labeling | F-score (%): 57.1 -> 35.9 | Hurt |
| add POS tag embeddings to M-BERT (ablation) | GATE · ACE05 En→Zh event argument role labeling | F-score (%): 52.5 -> 49.3 | Hurt |
| add dependency relation label embeddings to M-BERT (ablation) | GATE · ACE05 En→Zh event argument role labeling | F-score (%): 52.5 -> 49.7 | Hurt |
| add entity type embeddings to M-BERT (ablation) | GATE · ACE05 En→Zh event argument role labeling | F-score (%): 52.5 -> 57.8 | Helped |
| add entity type embeddings to M-BERT (ablation) | GATE · ACE05 En→Ar event argument role labeling | F-score (%): 47.4 -> 60.2 | Helped |
| drop directional information from relative structural position encoding (ablation) | relative structural position baseline · ACE05 En→Ar event argument role labeling | F-score (%): 47.1 -> 52.2 | Helped |
Closest papers by meaning
- Relation Extraction with Convolutional Network over Learnable Syntax-Transport Graph (AAAI 2020 · similarity 0.88)
- Event Detection via Gated Multilingual Attention Mechanism (AAAI 2018 · similarity 0.87)
- Dependency Structure-Enhanced Graph Attention Networks for Event Detection (AAAI 2024 · similarity 0.87)
- BattRAE: Bidimensional Attention-Based Recursive Autoencoders for Learning Bilingual Phrase Embeddings (AAAI 2017 · similarity 0.87)
- Dynamic Syntactic Feature Filtering and Injecting Networks for Cross-lingual Dependency Parsing (AAAI 2025 · similarity 0.85)
- Attention Mechanisms Perspective: Exploring LLM Processing of Graph-Structured Data (ICML 2025 · 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.