Low friction FAIR interoperability using RO-Crate metadata in text analytics pipelines

Research Object Crate (RO-Crate) is a simple method for linked-data description and packaging of data. Since 2021, the Language Data Commons of Australia (LDaCA) project has onboarded a number of language data collections with thousands of files. These are all consistently described as RO-Crates using a human and machine-readable Metadata Profile, discoverable through an online portal, and available via an access-controlled API. This presentation will show how analytics workflows can be connected to data in the LDaCA repository and use linked data descriptions, such as the W3C CSV for the web (CSVW) standard, to automatically detect and load the right data for analytical workflows. We will show how the general-purpose flexible linked metadata and raw data is prepared for use with common tools implemented in Jupyter notebooks. This work, funded by the Australian Research Data Commons ARDC, has enabled novel research by making data collected using sub-disciplinary norms of linguistics available to researchers working in other specialised areas - we will show examples of this and how this approach is relevant to other HASS and STEM disciplines, demonstrating work which would not have been possible without this co-investment between the Language Data Commons partners and ARDC The presentation should be accessible to the broad audience of eResearch and be of particular relevance to those with an interest in workflows and analytics, as well as metadata, vocabulary and repository specialists. It shows a FAIR research system which runs on open specifications and code and can be redeployed for other domains.