OCR'd Prefaces & Front Matter
Most CDSL dictionaries are scholarly works whose front matter — title pages, forewords (Vorwort / Vorrede / preface), lists of cited works and abbreviations, and addenda/corrigenda — contains the editor's own account of sources, method, and conventions. These pages are scanned in csl-doc but, until now, not available as searchable text or in translation.
This page is the public index of OCR'd front matter: each dictionary row links the
in-repo prefaces/ README and the consolidated single-file editions.
- Per-page files —
…prefNN.md(faithful transcription in the source language), plus…prefNN.en.md/…prefNN.ru.md(translations) when needed. - Consolidated editions — one file per language with all pages in order and a table of
contents (
…pref_all.de.md,…pref_all.en.md,…pref_all.ru.md, …). - Original scans under
prefaces/scans/where the source is multi-page PNG/JPG scans. - Operator manual (phases, engines A/B, attribution): Preface OCR pipeline.
- Sample quality probe (PWG/PW, H1559): OCR + EN error-rate report — stratified 15 pages, scan-locus + dual-pass + EN sense tables.
Sanskrit (Devanāgarī and transliteration), work titles, bibliographic abbreviation keys, reference numbers, and Latin quotations are kept verbatim across languages; digitizer running-header/footer stamps are omitted.
How this is produced: /cologne-preface-ocr — vision band OCR on csldoc scan pages, or
HTML text-layer extract when csldoc already has a transcribed single page (AP90, AE, BEN, ACC).
A separate Tesseract pipeline in PD (feat/ocr-v2-pipeline)
covers Deccan College Encyclopaedic Dictionary PDF front matter — not CDSL.
Cologne scan index (all codes that have front-matter pages): csldoc dictionaries.
Available now
Census 23-07-2026 (post-merge of AP90 #49, BEN #32, ACC #25, BUR #13). Page counts are source-language pages (approx.).
Queue (incomplete)
| Priority | Codes | Notes |
|---|---|---|
| Partial | PE (genealogy tables 05–26) | Title / foreword / preface done |
| Multi-scan cold | SKD (~32), PUI (~37) | csldoc PNG toctrees |
| No usable pref HTML | AP, ABCH, LRV | 404 / missing toctree |
Operator manual: Preface OCR pipeline.
What's covered (highlights)
- PWG / PW / GRA / SCH / CCS — German St Petersburg / Grassmann / Schmidt / Cappeller.
- MW family — Monier-Williams SE, ES, 1872.
- Apte / Benfey / ACC — HTML text-layer extracts (csldoc already transcribed).
- BUR — full Burnouf French front matter (28 pages) with EN + RU.
- LAN / GST / PGN / SNP / AE — staging repos with complete editions.
- VCP / KRM / BOP / STC — Sanskrit / Latin / French indigenous and European sources.
More dictionaries will be added as their front matter is transcribed via /cologne-preface-ocr.