Falls Church History Room Index & IIIFScribe

When I discovered that the Mary Riley Styles Public Library kept its five local-history indexes online only as scanned images, I knew there had to be a better way to search them.

The Problem

Hundreds of pages of priceless Falls Church history were locked inside images — no text search, no filtering, and no easy way for historians (or curious locals like me) to explore the collection.

My Solution

1. IIIFScribe — a custom OCR pipeline that:
   • Accepts any IIIF manifest
   • Downloads each high-resolution page
   • Extracts text with Tesseract.js
   • Generates a markdown transcript with full front-matter and page markers
2. Data Pipeline
   • Converts the transcript to JSON
   • Loads it into a Supabase Postgres table
3. Search Interface
   • Built with Next.js 14 App Router + shadcn/ui
   • Boolean (AND/OR) multi-term search across title, subject, notes, and more
   • Downloads results as JSON
4. AI-Powered Suggestions
   • An OpenAI-backed endpoint that turns a research prompt into optimized search terms tailored to the Falls Church corpus.

What I Like

• First time wrangling IIIF manifests at scale — now it feels easy.
• Watching the OCR progress bar chew through 700+ pages in real time.
• The AI suggestion panel routinely surfaces terms I never would have thought of.