Research map / Large language models / Retrieval and knowledge
K-BERT: Enabling Language Representation with Knowledge Graph
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 |
|---|---|---|---|
| Inject HowNet triples | K-BERT · Chnsenticorp test | accuracy: 94.3 -> 95.6 | Helped |
| Inject HowNet triples | K-BERT · Book_review, Shopping, and Weibo | accuracy: 87.5 -> 87.2; 96.3 -> 97.0; 98.3 -> 98.3 | No clear effect |
| Use HowNet rather than CN-DBpedia | K-BERT · XNLI and LCQMC | accuracy: XNLI 75.4 -> 76.1; LCQMC 86.2 -> 86.9 | Mixed |
| Inject CN-DBpedia triples | K-BERT · NLPCC-DBQA and MSRA-NER test | MRR / F1: 93.3 -> 94.3; 93.6 -> 95.7 | Helped |
| Inject domain KG knowledge | K-BERT · Finance_Q&A and Law_Q&A test | F1: 83.9 -> 84.9; 86.4 -> 87.5 | Helped |
| Inject MedicalKG triples | K-BERT · Medicine_NER test | F1: 92.5 -> 94.2 | Helped |
| Add soft-position embeddings (ablation) | K-BERT with KG · Law_Q&A and Medicine_NER | F1 | Helped |
| Add visible matrix to constrain attention (ablation) | K-BERT with KG · Law_Q&A and Medicine_NER | F1 | Helped |
| Allow all tokens to attend to each other (ablation) | K-BERT with CN-DBpedia · Law_Q&A | F1 | Hurt |
| Add WebtextZh to pre-training | BERT · MSRA-NER test | F1: 93.6 -> 94.6 | Helped |
Closest papers by meaning
- K-ON: Stacking Knowledge on the Head Layer of Large Language Model (AAAI 2025 · similarity 0.88)
- A Compact Model for Mathematics Problem Representations Distilled from BERT (AAAI 2025 · similarity 0.87)
- MKGL: Mastery of a Three-Word Language (NeurIPS 2024 · similarity 0.87)
- Select and Augment: Enhanced Dense Retrieval Knowledge Graph Augmentation (Abstract Reprint) (AAAI 2024 · similarity 0.86)
- KGGen: Extracting Knowledge Graphs from Plain Text with Language Models (NeurIPS 2025 · similarity 0.86)
- Editing Language Model-Based Knowledge Graph Embeddings (AAAI 2024 · similarity 0.86)
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.