Research map / Architectures and efficiency / Attention and transformers
A Lightweight Neural Model for Biomedical Entity Linking
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 lightweight alignment-based ranking model instead of BERT-based Ranking | Biomedical entity linking · ShARe/CLEF, NCBI, ADR | accuracy: 90.10$\pm$1.00 vs 91.06$\pm$0.96; 89.07$\pm$2.63 vs 89.06$\pm$2.63; 92.63$\pm$0.81 vs 93.22$\pm$0.79 | No clear effect |
| Add character-level features (ablation) | Our Base Model · ShARe/CLEF, NCBI, ADR | accuracy: -1.21; -0.31; -0.30 | Helped |
| Add the alignment layer (ablation) | Our Base Model · ShARe/CLEF, NCBI, ADR | accuracy: -3.80; -4.06; -3.17 | Helped |
| Add the CNN layer (ablation) | Our Base Model · ShARe/CLEF, NCBI, ADR | accuracy: -1.87; -0.93; -0.35 | Helped |
| Add mention-entity prior (ablation) | Our Base Model · ShARe/CLEF, NCBI, ADR | accuracy change: +0.33; +0.04; +0.03 | Helped |
| Add context features (ablation) | Our Base Model · ShARe/CLEF, NCBI, ADR | accuracy change: -0.09; +0.21; -0.24 | Mixed |
| Add entity-coherence features (ablation) | Our Base Model · ShARe/CLEF, NCBI, ADR | accuracy change: -0.02; +0.27; +0.11 | Helped |
| Change word ordering in mention names | Our Base Model · Simulated ADR datasets | accuracy: 92.63 -> 92.20; 92.63 -> 92.18; 92.63 -> 91.95; 92.63 -> 92.31; 92.63 -> 92.05 | No clear effect |
| Add typos to mention names | Our Base Model · Simulated ADR datasets | accuracy: 92.63 -> 92.03; 92.63 -> 91.61; 92.63 -> 91.38; 92.63 -> 91.41; 92.63 -> 91.13 | Hurt |
| Use the alignment-based lightweight model instead of BERT models | Biomedical entity linking · Average inference time | inference speed: 6.4x faster | Helped |
Closest papers by meaning
- Improving Biomedical Entity Linking with Cross-Entity Interaction (AAAI 2023 · similarity 0.91)
- LATTE: Latent Type Modeling for Biomedical Entity Linking (AAAI 2020 · similarity 0.91)
- Simultaneously Linking Entities and Extracting Relations from Biomedical Text without Mention-Level Supervision (AAAI 2020 · similarity 0.89)
- Association Pattern-aware Fusion for Biological Entity Relationship Prediction (NeurIPS 2024 · similarity 0.82)
- MEPNet: Medical Entity-Balanced Prompting Network for Brain CT Report Generation (AAAI 2025 · similarity 0.82)
- A Dual-Way Enhanced Framework from Text Matching Point of View for Multimodal Entity Linking (AAAI 2024 · similarity 0.82)
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.