Research map / Label-efficient and robust learning / Recognition and long-tail
Exploring Self-Distillation Based Relational Reasoning Training for Document-Level Relation 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 the four-pattern R-MSA reasoning module and self-distillation training instead of the ATLOP-BERT setup | Document-level relation extraction with BERT-base · DocRED test | F1: 61.30 -> 62.85 | Helped |
| Use self-distillation rather than vanilla knowledge distillation (ablation) | Ours-BERT · DocRED development | F1: 61.76 -> 62.81 | Helped |
| Use self-distillation rather than R-Drop (ablation) | Ours-BERT · DocRED development | F1: 61.92 -> 62.81 | Helped |
| Replace standard MSA with the four-pattern R-MSA unit (ablation) | Ours-BERT · DocRED development | F1: 59.39 -> 62.81 | Helped |
| Add the second, third, and fourth reasoning patterns (ablation) | Ours-BERT reasoning module · DocRED development | F1: 62.16 -> 62.81 | Helped |
| Add the masked entity-pair training branch (ablation) | Ours-BERT self-distillation framework · DocRED development | F1: 61.53 -> 62.81 | Helped |
| Add the unmasked supervised training branch (ablation) | Ours-BERT self-distillation framework · DocRED development | F1: 62.38 -> 62.81 | Helped |
| Add curriculum learning for selecting masked entity pairs (ablation) | Ours-BERT self-distillation framework · DocRED development | F1: 62.56 -> 62.81 | Helped |
| Use the proposed model instead of DocuNet-SciBERT | Document-level relation extraction with SciBERT-base · CDR test | F1: 76.3 -> 76.8 | Helped |
| Use the proposed model instead of DocuNet-SciBERT | Document-level relation extraction with SciBERT-base · GDA test | F1: 85.3 -> 86.4 | Helped |
Closest papers by meaning
- Revisiting Document-Level Relation Extraction with Context-Guided Link Prediction (AAAI 2024 · similarity 0.89)
- Self-Supervised Relational Reasoning for Representation Learning (NeurIPS 2020 · similarity 0.88)
- COMM: Concentrated Margin Maximization for Robust Document-Level Relation Extraction (AAAI 2025 · similarity 0.88)
- DocKS-RAG: Optimizing Document-Level Relation Extraction through LLM-Enhanced Hybrid Prompt Tuning (ICML 2025 · similarity 0.88)
- Enhancing Relation Extraction via Supervised Rationale Verification and Feedback (AAAI 2025 · similarity 0.85)
- A Hierarchical Network for Multimodal Document-Level Relation Extraction (AAAI 2024 · 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.