silvertext.
All posts

Category1 post

Engineering

Write-ups and measurements from building the checker: methods, numbers, limitations.

A side view in a preservation lab of an open bound volume in a V-shaped scanner cradle beneath two cameras, with gloved hands turning a page.

Engineeringengineering6 min read

How OCR Errors Change Document Similarity Before Matching Begins

An engineering review of how OCR, PDF extraction, reading order and hyphenation change text before similarity matching begins.