Research map / Vision-language and multimodal / Visual question answering
Hybrid-DMKG: A Hybrid Reasoning Framework over Dynamic Multimodal Knowledge Graphs for Multimodal Multihop QA with Knowledge Editing
Li Yuan, Qingfei Huang, Bingshan Zhu, Yi Cai, Qingbao Huang, Changmeng Zheng, Zikun Deng, Tao Wang
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 relation-linking prediction (ablation) | Hybrid-DMKG · MMQAKE (original image), BLIP-2 | H-Acc (%): 18.59 -> 28.88 | Improved |
| add relation-linking prediction (ablation) | Hybrid-DMKG · MMQAKE (rephrased image), MiniGPT-4 | H-Acc (%): 8.30 -> 22.23 | Improved |
| add RAG-enhanced reasoning in LVLM (ablation) | Hybrid-DMKG · MMQAKE (original image) | H-Acc (%): 21.50 -> 28.88 | Improved |
| add RAG-enhanced reasoning in LVLM (ablation) | Hybrid-DMKG · MMQAKE (rephrased image) | H-Acc (%): 19.41 -> 26.08 | Improved |
| add background-reflective decision (ablation) | Hybrid-DMKG · MMQAKE (original image), MiniGPT-4 | H-Acc (%): 20.23 -> 24.73 | Improved |
| add background-reflective decision (ablation) | Hybrid-DMKG · MMQAKE (rephrased image), MiniGPT-4 | H-Acc (%): 16.58 -> 22.23 | Improved |
| add background-reflective decision (ablation) | Hybrid-DMKG · MMQAKE (original image), BLIP-2 | M-Acc (%): M-Acc 48.19 -> 47.55 | No clear effect |
| replace Gemini decomposer with LLaMA2-7B (ablation) | Hybrid-DMKG · MMQAKE (original image), BLIP-2 | H-Acc (%): 28.88 -> 23.39 | Hurt |
| replace Gemini decomposer with ChatGPT (ablation) | Hybrid-DMKG · MMQAKE (original image), BLIP-2 | H-Acc (%): 28.88 -> 27.08 | Hurt |
| replace CLIP retriever with MM-Embed (ablation) | Hybrid-DMKG · MMQAKE (original image), BLIP-2 | H-Acc (%): 28.88 -> 26.12 | Hurt |
| replace CLIP retriever with MM-Embed (ablation) | Hybrid-DMKG · MMQAKE (original image), LLaVA | H-Acc (%): 29.90 -> 30.67 | Mixed |
| remove alias set from evaluation | Hybrid-DMKG evaluation · MMQAKE (original image), LLaVA | H-Acc (%): 29.90 -> 12.99 | Hurt |
Closest papers by meaning
- Benchmarking Multimodal Retrieval Augmented Generation with Dynamic VQA Dataset and Self-adaptive Planning Agent (ICLR 2025 · similarity 0.89)
- Augmenting Multimodal LLMs with Self-Reflective Tokens for Knowledge-based Visual Question Answering (CVPR 2025 · similarity 0.88)
- MDK12-Bench: A Multi-Discipline Benchmark for Evaluating Reasoning in Multimodal Large Language Models (AAAI 2026 · similarity 0.88)
- MME-CoT: Benchmarking Chain-of-Thought in Large Multimodal Models for Reasoning Quality, Robustness, and Efficiency (ICML 2025 · similarity 0.88)
- KAM-CoT: Knowledge Augmented Multimodal Chain-of-Thoughts Reasoning (AAAI 2024 · similarity 0.88)
- Benchmarking Retrieval-Augmented Multimomal Generation for Document Question Answering (NeurIPS 2025 · similarity 0.88)
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.
