About
The Transcript Archive
The Transcript Archive preserves decades of student reporting, campus life, and community history. Browse by date, explore sections, and read the original layouts as they appeared in print.
This project celebrates the voices that shaped the paper and makes the archive accessible for students, alumni, and researchers.
How it was built
From bulk scans to a searchable, AI-augmented archive
Ohio Wesleyan University’s student newspaper, The Transcript, has been published since 1867. Decades of its print editions existed only as bulk scanned TIF files in the OCLC ContentDM archive — unsearchable, unstructured, and effectively invisible to anyone who didn’t already know the exact date they were looking for. The goal of this project was to turn half a century of that print history (1950–2006) into a tool anyone can query in plain language.
A Python OCR pipeline converts each raw scan into structured articles, advertisements, and cropped images. A TypeScript retrieval system then indexes that content as 768-dimension multimodal embeddings in Postgres with pgvector, and answers questions using hybrid search — combining vector similarity with full-text relevance — an agent loop for more complex queries, and answers that are grounded in cited source articles rather than invented.
Today the archive holds 351 fully ingested editions (1950–2006), 11,705 articles with complete multimodal embedding coverage, and 6,846 advertisements, alongside period-accurate weather (1950–2000) and US music-chart (1958–2010) sidebars that place each edition in its moment. Retrieval is designed around versioned, immutable index builds so results stay reproducible as the corpus grows.
The stack: Next.js 16 and React 19 on the front end; TypeScript throughout; a Python 3.12 OCR pipeline; Neon Postgres with pgvector for storage and retrieval; Cloudflare R2 for image assets; Google Gemini for embeddings and generation; and Vercel for deployment. Deeper write-ups of each subsystem live in the architecture docs.
Who built this
Mostafa Anwari
I designed and built The Transcript Archive end to end — the OCR and data pipelines, the retrieval system, and the reading experience — to make a piece of Ohio Wesleyan’s history genuinely usable again. I’m happy to talk through any part of how it works.