Configuring Zotero 7–9 Translation Pipelines, LaTeX Preservation & Terminology Glossaries

This technical guide explains how zotero-geyi-translator handles multi-column academic PDF layouts, mathematical notation, and scanned historical documents across supported languages.

1. Two-Column Layout Reconstruction & LaTeX Equation Preservation

Standard PDF translation tools often flatten two-column academic papers or corrupt inline mathematical symbols, subscripts, and numbered equation environments. When submitting a PDF from Zotero:

Structural Block Detection: Page regions are segmented into body paragraphs, two-column reading flows, figure/table captions, footnotes, and mathematical formula blocks.
Formula Isolation: Inline symbols ( $\alpha, \beta, \nabla$ ) and display equations are preserved in place while surrounding prose is translated into the target language.
Side-by-Side Bilingual Output: Selecting the Bilingual (双语对照) output mode generates a synchronized side-by-side PDF attachment inside Zotero so researchers can verify equations and citations against the original source page (Zotero Plugin Documentation).

2. Handling 100–500 Page Monographs vs. Journal Articles

Journal Articles (< 100 pages): Processed via the real-time translation pipeline for rapid turnaround during literature reviews (Academic Paper Translation).
Academic Books & Dissertations ($\ge 100$ pages): Automatically routed to the asynchronous monograph pipeline (Full-Book Translation), which enforces unified terminology and chapter-level layout continuity across up to 500 pages.

3. Scanned Archival PDFs & Structured Markdown Extraction

For non-born-digital PDFs (such as scanned archival holdings, vertical East Asian scripts, or historical European typefaces):

Enable the high-precision OCR pipeline to extract structured text before translation, or convert complex scientific PDFs directly into clean Markdown and LaTeX source files via Scientific PDF to Markdown.