Skip to content
Use case

Converting a two-column paper to EPUB

Updated:

Two-column papers break when a converter reads straight across both columns. On the OLMo paper (arXiv, CC BY 4.0) the EPUBLab engine keeps the columns apart, but pdf_oxide's extract_text scores higher on continuity for this document.

Original and result

Two columns per page, tables, figures and footnotes. Reading across the column gap mixes two unrelated sentences on every line.

Original PDF
Page of the original PDF
Result (EPUB, as a reader shows it)
The same passage in the converted EPUB

Download source PDF Download the sample EPUB

Source and licence

Document
OLMo paper (arXiv 2402.00838)
Licence
CC BY 4.0
Pages
21
Traits
text PDF, two columns, tables, footnotes, figures, academic paper

Settings used

  • Engine: epublab-core, build of 2026-11-03 (the one measured in the benchmark).
  • Pages: the whole document.
  • Options: none. The engine has no per-document settings yet; the same defaults apply to every file.

Validation and results

EPUBCheck EPUBCheck on our EPUB: 0 errors, 0 warnings.

8-gram recall · Like 4-gram recall with 8 words: drops when reading order or line joins break sentences (continuity). Methodology
Engine8-gram recall
EPUBLab engine (EPUB output)0.800
EPUBLab engine (plain text)0.800
pdf_oxide · extract_text0.854
pdf_oxide · to_plain_text (XY-cut)0.273
pdf_oxide · Markdown0.719
PDFium · text().all()0.761
OpenDataLoader PDF (local mode)0.820
pdftotext (Poppler)0.781

Limitations we observed

  • pdf_oxide extract_text has a higher 8-gram recall than our engine on this document.
  • Tables are written as text in reading order; they are not rebuilt as tables.

Convert a similar PDF