Research map / Architectures and efficiency / Attention and transformers
Self-Attention Enhanced Selective Gate with Entity-Aware Embedding for Distantly Supervised 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 SeG instead of PCNN+ATT | NYT relation extraction · NYT held-out test, all bags | P@N mean (%): 72.2 -> 89.3 | Helped |
| use SeG instead of PCNN+BAG-ATT | NYT relation extraction · NYT held-out test, all bags | P@N mean (%): 84.8 -> 89.3 | Helped |
| use SeG instead of PCNN+HATT | NYT relation extraction · NYT held-out test | AUC: 0.42 -> 0.51 | Helped |
| use SeG instead of PCNN+ATT | NYT relation extraction · NYT one-sentence bags | AUC: 0.35 -> 0.48 | Helped |
| add entity-aware embedding (ablation) | SeG · NYT held-out test, all bags | P@N mean (%): 79.0 -> 89.3 | Helped |
| add entity-aware embedding (ablation) | SeG · NYT held-out test | AUC: 0.40 -> 0.51 | Helped |
| add selective gate aggregation (ablation) | SeG · NYT held-out test, all bags | P@N mean (%): 87.7 -> 89.3 | Helped |
| add selective gate aggregation (ablation) | SeG · NYT held-out test | AUC: 0.48 -> 0.51 | Helped |
| add self-attention alongside PCNN and selective gate (ablation) | SeG · NYT held-out test | AUC: 0.47 -> 0.51 | Helped |
| replace mean pooling with selective attention (ablation) | SeG · NYT held-out test, all bags | P@N mean (%): 89.3 -> 83.6 | Hurt |
| replace selective gate aggregation with selective attention (ablation) | SeG · NYT held-out test | AUC: 0.51 -> 0.47 | Hurt |
| stack PCNN and self-attention instead of arranging them in parallel (ablation) | SeG · NYT held-out test, all bags | P@N mean (%): 89.3 -> 89.1 | Hurt |
Closest papers by meaning
- SEE: Syntax-Aware Entity Embedding for Neural Relation Extraction (AAAI 2018 · similarity 0.93)
- Cross-Relation Cross-Bag Attention for Distantly-Supervised Relation Extraction (AAAI 2019 · similarity 0.91)
- Distant Supervision for Relation Extraction with Sentence-Level Attention and Entity Descriptions (AAAI 2017 · similarity 0.89)
- Enhancing Low-Resource Relation Representations through Multi-View Decoupling (AAAI 2024 · similarity 0.87)
- DocKS-RAG: Optimizing Document-Level Relation Extraction through LLM-Enhanced Hybrid Prompt Tuning (ICML 2025 · similarity 0.86)
- A Positive-Unlabeled Metric Learning Framework for Document-Level Relation Extraction with Incomplete Labeling (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.