Every archived record an RO-Crate: A storage contract for enrichment on Fedora 6/7

Abstract

The Archaeological Map of the Czech Republic (AMCR) is the national information system for archaeological fieldwork. It runs on Fedora 6.5 (F7 just being implemented), and each record is one Archival Group, so one OCFL object. In the Horizon Europe project ATRIUM we are adding machine enrichment of the documents it holds: page types, OCR quality, named entities, keywords and English metadata. To keep the results reusable and traceable, we agreed a storage contract before writing the code. The talk presents that contract as upcoming work. Each enrichment output is either a named distribution beside the original file or a block of the file's enrichment record, itself a distribution; every distribution has its own paradata as an RO-Crate Process Run Crate `CreateAction`. Every archived record carries `ro-crate-metadata.json` at its root. The head version of its OCFL object is then an Attached RO-Crate 1.2 that can be read without Fedora or AMČR, and the same document, served through the File API, resolves to live URLs.