Research map / Graph learning / Knowledge and temporal graphs
Document-Level Relation Extraction with Reconstruction
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 path reconstruction during training | HeterGSAN · DocRED test | F1: 53.52 -> 55.23 | Helped |
| Add BERT and path reconstruction | HeterGSAN · DocRED test | F1: 58.54 -> 59.45 | Helped |
| Add the reconstruction signal at inference after training with reconstruction (ablation) | HeterGSAN · DocRED dev | F1: 55.66 -> 56.22 | Helped |
| Add reconstruction training without using reconstruction at inference (ablation) | HeterGSAN · DocRED dev | F1: 54.40 -> 55.66 | Helped |
| Reconstruct paths for no-relation pairs as well as relation pairs (ablation) | HeterGSAN+Reconstruction · DocRED dev | F1: 55.55 -> 56.22 | Helped |
| Reconstruct paths only for no-relation entity pairs (ablation) | HeterGSAN · DocRED dev | F1: 54.40 -> 55.04 | Helped |
| Reconstruct paths only for relation entity pairs (ablation) | HeterGSAN · DocRED dev | F1: 54.40 -> 55.55 | Helped |
| Reconstruct Meta Path 1 only (ablation) | HeterGSAN · DocRED dev | F1: 54.40 -> 54.79 | Helped |
| Reconstruct Meta Path 2 only (ablation) | HeterGSAN · DocRED dev | F1: 54.40 -> 54.78 | Helped |
| Reconstruct Meta Path 3 only (ablation) | HeterGSAN · DocRED dev | F1: 54.40 -> 54.54 | Helped |
| Increase inference weight on reconstructed path probability up to the reported optimum (ablation) | HeterGSAN+Reconstruction · DocRED dev | F1 | Mixed |
| Combine BERT with HeterGSAN | HeterGSAN · DocRED test | F1: 53.52 -> 58.54 | Helped |
Closest papers by meaning
- COMM: Concentrated Margin Maximization for Robust Document-Level Relation Extraction (AAAI 2025 · similarity 0.90)
- Revisiting Document-Level Relation Extraction with Context-Guided Link Prediction (AAAI 2024 · similarity 0.89)
- DocKS-RAG: Optimizing Document-Level Relation Extraction through LLM-Enhanced Hybrid Prompt Tuning (ICML 2025 · similarity 0.87)
- A Positive-Unlabeled Metric Learning Framework for Document-Level Relation Extraction with Incomplete Labeling (AAAI 2024 · similarity 0.87)
- An Autoregressive Text-to-Graph Framework for Joint Entity and Relation Extraction (AAAI 2024 · similarity 0.86)
- 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.