Convert Double-Column Academic Papers to Markdown with Docling
Stop fighting broken multi-column layouts and scrambled LaTeX equations. Learn how researchers use Docling and SheetParser to import arXiv papers into Obsidian and Notion.
Need to Convert an arXiv Paper Right Now?
Use the dedicated Academic Paper to Markdown tool on SheetParser. 100% free with KaTeX preview.
The Challenge with Academic Scientific Papers
Academic papers from conferences like NeurIPS, CVPR, ICML, and journals published on IEEE, ACM, and arXiv utilize a standard two-column layout.
When an ordinary parser reads these PDFs, it reads left-to-right across the entire page geometry. Line 1 of column A is immediately followed by Line 1 of column B. The result is completely garbled:
Furthermore, mathematical equations, matrices, and greek symbols like α, β, and Σ turn into corrupted unicode characters (e.g. "?").
How Docling Solves Double-Column Reading Order
IBM Docling utilizes a vision-based layout segmentation pipeline trained on DocLayNet.
- Column Bounding: The model isolates the left and right columns visually before any text is extracted.
- Reading Order Flow: A directed topological graph reads column 1 completely before proceeding to column 2.
- Footnote & Figure Captions: Captions attached to charts and tables are preserved near their visual context rather than breaking paragraph flow.
- Native KaTeX / LaTeX: Formula regions are routed to a specialized equation OCR network that outputs valid LaTeX code.
Exporting to Obsidian, Notion & Overleaf
Once converted via SheetParser's Academic Converter, your paper is ready for:
Obsidian
Save as .md. Math formulas render immediately using MathJax/KaTeX.
Notion
Paste Markdown directly. Math blocks are converted to native Notion equation blocks automatically.
Overleaf & LaTeX
Copy extracted LaTeX equations directly into your own TeX source files for citations.