GlossAPI Documentation

Welcome to the refreshed docs for GlossAPI, the GFOSS pipeline for turning academic PDFs into clean Markdown and metadata artifacts.

Start here

• Onboarding Guide — prerequisites, install choices, and first run.
• Quickstart Recipes — common extraction/OCR flows in copy-paste form.
• Lightweight PDF Corpus — 20 one-page PDFs for smoke testing without Docling or GPUs.

Learn the pipeline

• Code Map links the main documentation ideas to the classes and files that implement them.
• Pipeline Overview explains each stage and the emitted artifacts.
• OCR & Math Enrichment covers DeepSeek OCR remediation and Docling-based enrichment.
• OCR Repetition Policy pins the default repetition thresholds for word and LaTeX cleaning.
• OCR Cleaning Runtime explains the shared clean/debug analyzer, ordering, and why the cleaner separates tables, numeric, LaTeX, hybrid, and text ownership.
• Multi-GPU & Benchmarking shares scaling and scheduling tips.

Configure and debug

• Configuration lists all environment knobs.
• Troubleshooting captures the most common pitfalls.
• AWS Job Distribution describes large-scale scheduling.

Reference

• Corpus API gives the compact contract view of the main public methods.
• Legacy Corpus API Notes remains available while the docs are being consolidated.