AI Paper Atlas

Research map / Vision-language and multimodal / Human and embodied modeling

M$^{3}$3D: A Multimodal, Multilingual and Multitask Dataset for Grounded Document-Level Information Extraction

Jiang Liu, Bobo Li, Xinran Yang, Na Yang, Hao Fei, Mingyao Zhang, Fei Li, Donghong Ji

TPAMI 2025 · Vision-language and multimodal › Human and embodied modeling · approach: Multimodal retrieval-augmented modeling · 0 citations (OpenAlex)

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 changedWhereMeasuredEffect
add MMCM (missing-modality compensation) (ablation)Ours · M3D English avgF1: 53.03 -> 53.80Improved
add MMCM (ablation)Ours · M3D Chinese avgF1: 53.11 -> 53.77Improved
add DFFM (hierarchical modality fusion) (ablation)Ours · M3D English avgF1: 52.78 -> 53.80Improved
add DFFM (ablation)Ours · M3D Chinese avgF1: 51.86 -> 53.77Improved
add DFFM (ablation)Ours · M3D Chinese relation extractionF1: 31.80 -> 36.89Improved
add MMCM (ablation)Ours · M3D English relation extractionF1: 28.17 -> 29.78Improved
add hierarchical fusion over strongest baseline MDocRE-HNMDocRE-HN · M3D English avgF1: 52.65 -> 53.80Improved
add hierarchical fusion over strongest baseline MDocRE-HNMDocRE-HN · M3D Chinese avgF1: 52.42 -> 53.77Improved
use Ours instead of ViViT+Longformer+TSViViT+Longformer+TS · M3D English entity recognitionF1: 92.03 -> 91.24Hurt
use zero-shot Video-ChatGPT instead of trained modelVideo-ChatGPT · M3D English avgF1: 53.80 -> 22.47Hurt
use two-stream instead of one-stream fusionViViT+Longformer · M3D English avgF1: 50.99 -> 52.66Improved
use ViViT instead of VideoMAE with OS fusionLongformer+OS · M3D Chinese avgF1: 47.78 -> 50.92Improved

Closest papers by meaning

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy