Research map / Graph learning / Knowledge and temporal graphs
Procedural Text Understanding via Scene-Wise Evolution
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 scene-wise SGR instead of the strongest reported entity-wise baseline | Procedural text understanding · ProPara document-level | F1: 70.5 -> 72.2 | Helped |
| Use scene-wise SGR instead of the strongest reported baseline | Procedural text understanding · ProPara sentence-level | Macro-Avg: 58.8 -> 59.5 | Helped |
| Use scene-wise SGR instead of KOALA | Procedural text understanding · Recipes re-split | F1: 56.1 -> 58.4 | Helped |
| Track entities scene-by-scene instead of entity-by-entity | ProPara inference · ProPara | Total inference time (seconds): 51.31 -> 17.99 | Helped |
| Add a Graph Structure Encoder (ablation) | SGR · ProPara document-level | F1: 59.9 -> 72.2 | Helped |
| Add a Context Encoder (ablation) | SGR · ProPara document-level | F1: 64.1 -> 72.2 | Helped |
| Add ConceptNet knowledge (ablation) | SGR · ProPara document-level | F1: 71.6 -> 72.2 | Helped |
| Use pre-trained BERT (ablation) | SGR · ProPara document-level | F1: 68.7 -> 72.2 | Helped |
| Use ConceptNet during testing while omitting it during training (ablation) | SGR · ProPara document-level | F1: 71.6 -> 71.8 | Helped |
| Use ConceptNet during training but omit it during testing (ablation) | SGR · ProPara document-level | F1: 71.6 -> 69.4 | Hurt |
| Use ConceptNet during training and testing (ablation) | SGR · ProPara document-level | F1: 71.6 -> 72.2 | Helped |
| Add ConceptNet (ablation) | SGR · ProPara document-level | Recall: 63.2 -> 62.9 | No clear effect |
Closest papers by meaning
- Scene Graph Thinking: Reinforcing Structured Visual Reasoning for Multimodal Large Language Models (ICML 2026 · similarity 0.85)
- Visual Document Understanding and Reasoning: A Multi-Agent Collaboration Framework with Agent-Wise Adaptive Test-Time Scaling (CVPR 2026 · similarity 0.85)
- Schema-Guided Scene-Graph Reasoning Based on Multi-Agent Large Language Model System (AAAI 2026 · similarity 0.85)
- DocHop: Benchmarking Out-of-domain Multi-hop Reasoning in Information-Dense Documents (ICML 2026 · similarity 0.84)
- DiGraphHal-Bench: Evaluating Multimodal Large Language Models on Complex Directed Graphs (CVPR 2026 · similarity 0.84)
- Scene Graph Generation with Role-Playing Large Language Models (NeurIPS 2024 · similarity 0.84)
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.