SciDiagramEdit: Learning to Edit Scientific Diagrams from Paper Revisions

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Yasheng Sun, Zezi Zeng, Yifan Yang, Chong Luo, Wenyi Wang, Ziwei Liu, Jürgen Schmidhuber
• Venue: arXiv: cs.CL
• Published: 2026-07-16
• Collected: 2026-07-17
• URL: http://arxiv.org/abs/2607.15272v1
• PDF: https://arxiv.org/pdf/2607.15272v1
• DOI: UNKNOWN

Plain-Language Brief

The text is about a new system called SciDiagramEdit that automates the editing of scientific diagrams in research papers based on natural language instructions. The system learns from revisions made to figures in arXiv version histories and operates on editable vector sources. It uses agentic learning via skill evolution, which refines the system's skills over multiple epochs to improve edit accuracy.

What the Document Says

• Scientific figure editing is a time-consuming part of research practice.
• Authors relabel components, rearrange panels, and restyle visuals in revisions.
• Automating this workflow under natural language instructions is challenging due to the dense nature of scientific figures.
• SciDiagramEdit is a framework that learns from paper revisions and operates on the figure's editable vector source.
• Users can inspect and co-edit individual primitives alongside the agent.
• The benchmark mines before/after figure pairs from arXiv version histories.
• The diversity of editing instructions is accommodated through agentic learning via skill evolution.
• The system's skills are refined over multiple epochs to improve edit accuracy.
• Natural paper revisions are shown to be an effective training signal for instruction-driven figure editing.

Key Concepts Mentioned

• scientific diagrams
• paper revisions
• natural language instructions
• automation
• agentic learning
• skill evolution
• figure editing

Claims Made by the Document

• Scientific figure editing is time-consuming.
• SciDiagramEdit is a framework that learns from paper revisions and operates on editable vector sources.
• Users can inspect and co-edit individual primitives alongside the agent.
• Natural paper revisions are an effective training signal for instruction-driven figure editing.

Evidence Used

• The text states that editing figures in research papers is a routine and time-consuming part of everyday research practice.
• The text describes SciDiagramEdit as a framework that learns from natural paper revisions and operates on the figure's editable vector source.
• The text mentions that users can inspect and co-edit individual primitives alongside the agent.
• The text provides evidence that natural paper revisions are an effective training signal for instruction-driven figure editing through the improvement in edit accuracy over multiple epochs.

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Editing the figures in a research paper is a routine and time-consuming part of everyday research practice: authors relabel components, rearrange panels, and restyle visuals as they revise their manuscripts. Automating this editing workflow under a natural-language instruction, however, is challenging, because a scientific figure is a dense infographic in which heterogeneous visual elements such as schematics, plots, photos, captions, and arrows are composed under a tight visual grammar to advance a specific argument. To address this, we present SciDiagramEdit, a benchmark and skill-evolution framework that learns from natural paper revisions and operates on the figure's editable vector source, where users can inspect and co-edit individual primitives alongside the agent. Our benchmark mines before/after figure pairs from arXiv version histories, each grounded in the authors' own revision intent. To accommodate the diversity of editing instructions, we adopt agentic learning via skill evolution: an agentic proposer continually refines the agent's skill specification from execution traces over multiple epochs. The resulting skill progressively lifts edit accuracy on a held-out validation set, providing evidence that natural paper revisions are an effective training signal for instruction-driven figure editing.