Research map / Graph learning / Knowledge and temporal graphs
Discovering Latent Facts from Context to Construct Richer Open Knowledge Graphs
Jinpeng Li, Hang Yu, Ziqi Ma, Peng Qi
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| wrap ChatGPT-4.0 in the KG-DLF pipeline (normalizer, discoverer, corrector) | ChatGPT-4.0 · REBEL* (1,000 long texts) | F1: 0.415 -> 0.639 | Improved |
| wrap ChatGPT-4.0 in the KG-DLF pipeline | ChatGPT-4.0 · WebNLG* test | F1: 0.720 -> 0.845 | Improved |
| wrap ChatGPT-3.5 in the KG-DLF pipeline | ChatGPT-3.5 · Wiki-NRE* (cross-text, 1,000 instances) | F1: 0.588 -> 0.800 | Improved |
| wrap LLaMA3-8B-Instruct in the KG-DLF pipeline | LLaMA3-8B-Instruct · Wiki-NRE* (cross-text, 1,000 instances) | F1: 0.616 -> 0.556 | Hurt |
| wrap Mistral-7B-Instruct in the KG-DLF pipeline | Mistral-7B-Instruct · WebNLG* test | F1: 0.216 -> 0.211 | No clear effect |
| add knowledge discovery and correction on top of extraction (full KG-DLF vs KE+KN) (ablation) | KG-DLF (ChatGPT-4.0) · WebNLG* test | F1: 0.842 -> 0.845 | No clear effect |
| add KD and KC to the pipeline (full model vs a two-component variant) (ablation) | KG-DLF (ChatGPT-4.0) · REBEL* (1,000 long texts) | F1: 0.618 -> 0.639 | Mixed |
| increase number of in-context examples to 5 (ablation) | knowledge extractor · WebNLG*, REBEL*, Wiki-NRE* | F1 | Improved |
| increase number of clues k from 3 to 4 (ablation) | knowledge discoverer · three datasets | F1: storage 1.5MB -> 3.3MB | No clear effect |
| raise similarity thresholds beyond lambda_E 0.80 and lambda_R 0.75 (ablation) | knowledge normalizer · REBEL*, Wiki-NRE* | F1 | Hurt |
Closest papers by meaning
- KGGen: Extracting Knowledge Graphs from Plain Text with Language Models (NeurIPS 2025 · similarity 0.89)
- KG-FIT: Knowledge Graph Fine-Tuning Upon Open-World Knowledge (NeurIPS 2024 · similarity 0.88)
- A Systematic Exploration of Knowledge Graph Alignment with Large Language Models in Retrieval Augmented Generation (AAAI 2025 · similarity 0.87)
- MKGL: Mastery of a Three-Word Language (NeurIPS 2024 · similarity 0.87)
- Graph-constrained Reasoning: Faithful Reasoning on Knowledge Graphs with Large Language Models (ICML 2025 · similarity 0.87)
- Think-on-Graph 2.0: Deep and Faithful Large Language Model Reasoning with Knowledge-guided Retrieval Augmented Generation (ICLR 2025 · similarity 0.87)
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.
