Overview
ISO 16684-4:2024 - Graphic technology - Extensible metadata platform (XMP) specification - Part 4: Use of XMP for semantic units defines how XMP can describe and reference fine-grained pieces of content called semantic units (SUs). The standard extends the XMP concepts in ISO 16684-1 so that metadata can target sub-file or cross-file content (characters, paragraphs, image areas, audio/video segments, datasets, IRIs). It adopts the Web Annotation Data Model “target” syntax and specifies XMP serialization requirements to improve interoperability for linked data, metadata exchange, and AI-driven content processing.
Key topics and technical requirements
- Semantic unit (SU): a user-defined, meaningfully separable piece of content that can span files, pages, images, audio/video, or remote resources.
- Targeting mechanism: SUs must enumerate one or more target resources as defined by the W3C Web Annotation Data Model (specificResource, source, selector).
- Selectors and segmenting: selectors are used to identify segments of a source (e.g., paragraph, image region, time range) - multiple selector forms are supported depending on media type.
- XMP serialization: requirements and guidance on XMP packet syntaxes for SUs, including JSON-LD serialization conventions (see ISO 16684-3).
- Required XMP keywords for SU packets: @id, @context, @vocab, @language, @type, and target (and its qualifiers).
- Interoperability focus: aligns XMP with Web Annotation vocabulary to allow consistent referencing, dereferencing, and cross-system metadata exchange.
- Use cases and examples: digitized books, image regions, multimedia fragments, and composite SUs combining multiple resource types.
Applications and who should use it
- Publishers and digital libraries: fine-grained descriptive and preservation metadata for chapters, figures, tables, and scanned materials.
- Archivists and repository managers: robust cross-file references and long-term metadata for digital preservation.
- Metadata specialists and standards architects: consistent XMP-based metadata models that integrate with linked data and JSON-LD.
- Developers and software vendors: implementers of content management systems, annotation tools, and multimedia platforms needing precise targeting and serialization.
- AI and data scientists: training data annotation, content segmentation, and semantic enrichment pipelines that leverage granular, machine-actionable metadata.
Related standards
- ISO 16684-1:2019 - XMP data model, serialization and core properties
- ISO 16684-3:2021 - JSON-LD serialization of XMP
- W3C Web Annotation Data Model (2017) - target and selector vocabulary
Keywords: ISO 16684-4:2024, XMP, semantic unit, Web Annotation Model, XMP serialization, JSON-LD, metadata, linked data, digital preservation, content reuse, AI.