Research map / Graph learning / Knowledge and temporal graphs
SCOPE and SCION: A Benchmark and an Auditable Reference Pipeline for Schema Induction and Fusion from Text
Miaobo Hu, Xiaobo Guo, Shuhao Hu, BoKun Wang, Rui Chen, Xin Wang, Jun Xiao, Daren Zha
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 |
|---|---|---|---|
| add candidate-space constraints and contract-based abstraction (SCION-lite vs ETA) | extract-then-aggregate baseline · SCOPE core suite | Literal F1: +0.0455 Literal F1 (0.7063 -> 0.7518) | Improved |
| add candidate-space constraints and contract-based abstraction (SCION-lite vs ETA) | extract-then-aggregate baseline · SCOPE core suite | Graph F1: 0.7629 -> 0.7888 | Improved |
| constrained pipeline instead of free LLM-only induction (SCION-lite vs LLM-only GPT-5.2) | GPT-5.2 Chat · SCOPE core suite | Continuous F1: 0.8549 -> 0.8909 | Improved |
| replace proprietary schema engineer with RL-trained Qwen3-8B (SCION-RL) | SCION-lite · SCOPE core suite | Graph F1: 0.7888 -> 0.8228 | Improved |
| add deterministic typed-edge completion to released source schemas (ablation) | released source schemas · SCOPE core suite | Continuous F1: +0.0179 Continuous F1 | Improved |
| evaluate against train-derived reachable target instead of full gold (ablation) | SCION-lite · SCOPE core suite | Literal recall: 0.7145 -> 0.8464 | Improved |
| use SCION-lite schema instead of ETA or released schema in a fixed extractor | downstream extractor · Downstream instance-level extraction | macro F1: 0.5633 -> 0.6800 | Improved |
| add schema fusion to SCION-lite | SCION-lite · Downstream instance-level extraction | macro F1: 0.6800 -> 0.6900 | Improved |
| swap GPT-5.2 Chat for DeepSeek-R1 in LLM-only induction | LLM-only baseline · SCOPE core suite | Literal F1: 0.6836 -> 0.5664 | Hurt |
| add structural checks to fusion matching (ablation) | fusion matcher · Fusion audit (5k pair budget) | mapping precision: 0.81 precision, 0.06 conflict rate | Improved |
Closest papers by meaning
- DiscoveryBench: Towards Data-Driven Discovery with Large Language Models (ICLR 2025 · similarity 0.85)
- SO-Bench: A Structural Output Evaluation of Multimodal LLM (CVPR 2026 · similarity 0.85)
- STAR: Boosting Low-Resource Information Extraction by Structure-to-Text Data Generation with Large Language Models (AAAI 2024 · similarity 0.84)
- KGGen: Extracting Knowledge Graphs from Plain Text with Language Models (NeurIPS 2025 · similarity 0.83)
- SciReason: A Controllable Benchmark for Scientific Reasoning in LLMs (NeurIPS 2026 · similarity 0.83)
- A Systematic Exploration of Knowledge Graph Alignment with Large Language Models in Retrieval Augmented Generation (AAAI 2025 · similarity 0.83)
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.
