DIVE-to-design: how a multi-agent workflow converts figure-centric literature into an ai-native hydrogen storage discovery engine

От DIVE к проектированию: как мультиагентный рабочий процесс превращает литературу, ориентированную на рисунки, в нативный для ИИ механизм поиска материалов для хранения водорода
Yuyang Hong, Xin Mao, Xin Mao
2026-03-27

DIVE frameworkDigHydinverse designmulti-agent extractionsolid-state hydrogen storage
A major bottleneck in artificial intelligence (AI)-driven materials discovery is not model architecture, but limited data accessibility: critical experimental knowledge remains locked in figures, heterogeneous reporting formats, and unstructured PDFs. A recent study by Li et al. addresses this challenge by introducing DIVE (Descriptive Interpretation of Visual Expression), a multi-agent extraction framework that transforms figure-centric scientific content into structured, machine-actionable data. Applied to solid-state hydrogen storage materials, DIVE demonstrates substantial extraction gains over conventional direct large language model (LLM) parsing, then scales to mine 4,053 publications (1972-2025) and build a > 30,000-entry database that powers a downstream inverse-design agent, DigHyd. This work offers a practical blueprint for moving from “LLM-assisted reading” to “AI-enabled discovery infrastructure”, linking literature mining, quality scoring, database construction, and target-driven candidate generation in a single workflow.
1
DIVE is a multi-agent framework that converts figure-centric, heterogeneous scientific literature into structured, machine-actionable materials data.
2
DIVE scales to 4,053 publications published between 1972 and 2025, producing a database containing more than 30,000 entries.
3
For solid-state hydrogen storage materials, DIVE achieves substantially higher extraction performance than conventional direct large language model parsing.
4
The resulting database supports DigHyd, a downstream inverse-design agent for target-driven hydrogen-storage candidate generation.
5
The workflow integrates literature mining, extraction quality scoring, database construction, and AI-guided materials discovery into a unified infrastructure.

solid-state hydrogen storage materials and their figure-centric scientific literature

the extraction, structuring, quality scoring, and downstream use of figure-derived experimental knowledge for AI-driven discovery and inverse design

Publication Details
Publication Date
2026-03-27
Journal
Publisher
ISSN
Cited by
9
Access Type
Author Information
Authors
Yuyang Hong
Xin Mao
Xin Mao
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%