How to build conference proceedings
from Word submissions

Collect PDFs, not Word files

The single biggest simplification: ask authors for camera-ready PDFs, not .docx files. A Word file re-flows on every machine that opens it: different fonts installed, different printer drivers, different Word versions all shift line breaks and page counts. The author's PDF export is the only version of the paper that looks the way the author approved it.

If your template is in Word, that is fine: authors write in Word, then export to PDF themselves. Give them two instructions:

• Export via File → Save As → PDF with "ISO 19005-1 compliant (PDF/A)" checked, or via a PDF printer that embeds all fonts.
• Check their own export before submitting, especially the equations. Word's PDF export can drop legacy Equation Editor objects without any error message. We wrote up how to detect blank equations separately, because it hit roughly one in eight papers in one volume we checked.

Collect metadata in a structured form

For every paper, you need a title, authors in order, and affiliations. If you plan to register DOIs, also collect ORCID iDs and email addresses for corresponding authors. Collect all of this when the paper is uploaded, not afterwards by email. Chasing 100 authors for missing metadata after the deadline costs weeks.

If your review ran through EasyChair, EDAS, or OpenConf, export the accepted-papers list as CSV; titles and author lists carry over. Expect to correct them anyway: authors change titles between abstract and camera-ready more often than you would think.

Check compliance before you assemble

Every paper that enters the volume should pass four checks: correct page size, correct margins, all fonts embedded, and page count within the limit. One A4 paper in a Letter-sized volume, or one paper with a non-embedded font, can reject the whole print job. The checks take seconds per paper with free tools. Commands and thresholds are in our PDF compliance guide.

Assemble the volume

With compliant PDFs and clean metadata, assembly is a mechanical job that LaTeX does well. The pdfpages package includes each paper, keeps a running page counter, and lets you stamp page numbers and headers over the included pages:

\documentclass{book}
\usepackage{pdfpages}

\begin{document}
\frontmatter
% title page, preface, committee lists, table of contents...

\mainmatter
\includepdf[pages=-, pagecommand={\thispagestyle{plain}}]{papers/0001.pdf}
\includepdf[pages=-, pagecommand={\thispagestyle{plain}}]{papers/0002.pdf}
% ...one line per paper
\end{document}

In practice you will not write those lines by hand for 100+ papers. Generate them from your metadata spreadsheet with a small script, in paper order. The same script can produce the table-of-contents entries, so TOC and body can never disagree about a page number.

Papers should start on odd (right-hand) pages if the volume will be printed. That means inserting a blank page after every paper with an odd page count. pdfpages handles this with the openright logic of the book class, but verify it: an even-page start in the middle of the volume is the kind of error nobody notices until the printed copies arrive.

Front matter, TOC, and author index

The table of contents, session dividers, and author index come from your metadata, not from the PDFs. Generate them. A hand-maintained TOC for 100 papers will be wrong by the third revision of the paper order.

The author index (author name → page numbers) is worth the extra script: it is the feature that makes a proceedings volume usable as a reference, and it is pure drudgery to build manually.

Stamp DOIs and page numbers

If you register DOIs, each paper's first page should carry its DOI. Since the papers are finished PDFs, this is an overlay operation, the same mechanism as the page numbers. Do it as the last step, after the paper order is frozen, because every reorder changes every page number after the insertion point.

What goes wrong

• Mixed page sizes. A4 and US Letter differ by 6 mm in width and 18 mm in height. Mixed sizes make headers and page numbers land in different places per paper. Catch at submission.
• Non-embedded fonts. The volume renders fine on your machine (which has the font) and wrong at the printer (which does not). pdffonts tells you in one second.
• Blank equations from Word exports. No tool warns about these; the content is absent. See the detection guide.
• Author-applied headers and footers. Some authors submit with their own page numbers or a draft watermark. These collide with your stamped numbering and need to be redacted before inclusion.
• Late paper swaps. Every swapped paper after page numbering means re-stamping everything after it and re-checking the TOC and index. Freeze the paper order before you start numbering.

Or let the pipeline do it

Ready.Pub, the proceedings tool in the texlode ecosystem, runs this entire process as one workflow: contributors upload through a link and see compliance results immediately, metadata lives in an editable table, the volume compiles live as you reorder papers, and DOI stamping plus Crossref deposit XML come out at the end. Details on the proceedings page.