A new archival data structure: PARADISEC's move to RO-Crate

Presented by Peter Sefton, Amanda Harris and Nick Thieberger at the Language Documentation and Archiving Conference, Berlin & Online, 4-6 Sept, 2024 Over the life of the Pacific and Regional Archive for Digital Sources in Endangered Cultures (PARADISEC) we have used a few different tools for describing items in the collection. Each had its benefits, but each was superseded and allowed its contents to be re-used in the subsequent system, but each separated the catalog from the items in the collecBon. In 2024, we have adopted an emerging specification, the Research Object Crate [1]. The advantage of using RO-Crate as the underlying storage format is that it allows the collecBon to be self-describing, not relying on a database for contextual information. This makes the data more sustainable and transportable. A major advantage of using RO-Crate for PARADISEC is that we can now index the text files in RO-Crate and search within items, allowing users to search text files, and also to search transcripts of media (in ELAN’s .eaf format), with transcripts linked to playable media. This result then resolves to just the selected segment in a media file, a major advance over the view provided by the previous catalog. RO-Crate provides Linked Data metadata for data resources using JSON-LD, and the main metadata vocabulary is Schema.org. Linked data as used in RO-Crate also allows for interrelationships between files to be described, for example showing that an ELAN file is an annotation of a video file, and potenBally which Bers in the file are transcriptions, translations etc. The advantages of using RO-Crate metadata also gives the project access to multiple tool implementations such as (soWware libraries, metadata editors and data-discovery services. etc) In this presentation we will outline the new data structure of our collecBon and demonstrate the improved access to items in the collecBon that it provides.