SciXplain: Structure-Aware Representation Learning for Scientific Method Diagrams
Published in International Workshop on Document Analysis Systems (DAS), 2026
SciXplain uses structural and geometric supervision to improve the understanding of relationships in scientific method diagrams.
Recommended citation: Zhicheng He, Zheyi Deng, Junhao Ren, Xiaoqing Lyu. (2026). "SciXplain: Structure-Aware Representation Learning for Scientific Method Diagrams." Document Analysis Systems (DAS 2026), Pages 474-490.
Download Paper
