Manual transcription → eScriptorium
Schechter Institutes & National Library of Israel (MiDRASH)
Within the ERC-funded MiDRASH project, the Schechter Institutes and the National Library of Israel set out to extract text from around 30,000 medieval Hebrew, Aramaic and Judeo-Arabic manuscripts, drawing heavily on the Cairo Geniza. Replacing fragmented manual efforts, the team used eScriptorium to run global 'transcribe-a-thons': Kraken models generated rough baseline transcriptions, crowdsourced volunteers and scholars corrected them, and those corrections were fed back to recursively improve the models. The open-source approach ensured both the resulting data and the underlying algorithms remained permanently public. A leading Israeli example of combining HTR with large-scale scholarly crowdsourcing.
Original source
AI is enlisted to rescue Jews' cultural heritage and give new life to ancient languages
Religion News Service
The archived copy opens a snapshot on the Internet Archive's Wayback Machine, preserved for when the original moves or disappears.
The Archive Migration Review summarises this story in its own words and links to the original source for verification. We are editorially independent and not affiliated with the institution or software project named above. Summaries are compiled in good faith from publicly available accounts; corrections are welcome.