Research map / Video / Video-language understanding
MetaphorVU: Towards Metaphorical Video Understanding
Zhuoqun Li, Boxi Cao, Guiping Jiang, Fangrui Lv, Ruotong Pan, Jianan Wang, Xiangyu Wu, Hongyu Lin, Yaojie Lu, Yong Du, Ruyin Jia, Liyan and 4 more
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 MetaphorBoost knowledge-graph mapping augmentation | Gemini-3-Pro · MetaphorVU-Bench | average score: 63.8 -> 66.1 | Improved |
| add MetaphorBoost knowledge-graph mapping augmentation | Qwen2.5-VL-7B-Instruct · MetaphorVU-Bench | average score: 33.8 -> 37.9 | Improved |
| add MetaphorBoost knowledge-graph mapping augmentation | Qwen3-VL-8B-Thinking · MetaphorVU-Bench | average score: 52.0 -> 55.9 | Improved |
| add external knowledge augmentation (ablation) | MetaphorBoost (Qwen3-VL-8B-Thinking) · MetaphorVU-Bench | average score: 53.4 -> 55.9 | Improved |
| use graph-structure augmentation instead of plain text (ablation) | MetaphorBoost (Qwen3-VL-8B-Thinking) · MetaphorVU-Bench | average score: 54.3 -> 55.9 | Improved |
| use metaphor-oriented knowledge instead of commonsense knowledge (ablation) | MetaphorBoost (Qwen3-VL-8B-Thinking) · MetaphorVU-Bench | average score: 52.5 -> 55.9 | Improved |
| add common connection in query (ablation) | MetaphorBoost (Qwen3-VL-8B-Thinking) · MetaphorVU-Bench | average score: 54.8 -> 55.9 | Improved |
| use 2 hops instead of 1 hop (ablation) | MetaphorBoost (Qwen3-VL-8B-Thinking) · MetaphorVU-Bench | average score: 54.5 -> 55.9 | Improved |
| retain 10 results instead of 5 (ablation) | MetaphorBoost (Qwen3-VL-8B-Thinking) · MetaphorVU-Bench | average score: 55.7 -> 55.9 | No clear effect |
| apply LTR inference-time scaling | Qwen3-VL-8B-Thinking · MetaphorVU-Bench | average score: 52.0 -> 44.5 | Hurt |
| apply ViTCoT inference-time scaling | Qwen3-VL-8B-Thinking · MetaphorVU-Bench | average score: 52.0 -> 46.2 | Hurt |
| add metaphor-tailored prompt engineering | Qwen3-VL-8B-Thinking · MetaphorVU-Bench | average score: 52.0 -> 52.4 | No clear effect |
Closest papers by meaning
- ViMU: Benchmarking Video Metaphorical Understanding (NeurIPS 2026 · similarity 0.89)
- MMBench-Video: A Long-Form Multi-Shot Benchmark for Holistic Video Understanding (NeurIPS 2024 · similarity 0.89)
- MMOU: A Massive Multi-Task Omni Understanding and Reasoning Benchmark for Long and Complex Real-World Videos (NeurIPS 2026 · similarity 0.87)
- Video-of-Thought: Step-by-Step Video Reasoning from Perception to Cognition (ICML 2024 · similarity 0.87)
- Needle In A Video Haystack: A Scalable Synthetic Evaluator for Video MLLMs (ICLR 2025 · similarity 0.86)
- VideoHallu: Evaluating and Mitigating Multi-modal Hallucinations on Synthetic Video Understanding (NeurIPS 2025 · 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.
