Ingest is the controlled process of bringing assets into the DAM and establishing their identity, provenance and metadata from the outset. In OAIS (ISO 14721) terms, external material arrives as a Submission Information Package (SIP) and is transformed into an Archival Information Package (AIP) for management and a Dissemination Information Package (DIP) for delivery. Getting ingest right is decisive: metadata captured at the point of entry is cheap and accurate, while retrofitting it later is expensive and error-prone.

Ingest as a controlled pipeline

  • Validate and characterise: confirm file format and integrity, run format identification (PRONOM / DROID), and generate a fixity checksum (e.g. SHA-256) recorded in PREMIS.
  • Normalise where policy requires: convert fragile or proprietary formats to sustainable ones on ingest.
  • Assign identifiers: mint a persistent, opaque identifier so the asset can be cited and linked independently of filename or location.
  • Capture provenance: record source, agent and event history as PREMIS events.

Metadata: the four types

Practitioners distinguish descriptive metadata (title, creator, subject - often Dublin Core, EAD or MODS), technical metadata (format, resolution, codec, colour space), administrative and rights metadata (licence, owner, retention), and structural metadata (how parts relate - pages in a book, tracks on a reel). Embedded standards such as IPTC, EXIF and XMP let some of this metadata live inside the file itself, which is invaluable when assets travel outside the DAM.

Taxonomy and controlled vocabulary

Free-text tagging drifts into synonyms and typos; controlled vocabularies keep retrieval consistent. Use a taxonomy (hierarchical categories) for browsing and controlled keyword lists or thesauri - Getty AAT, TGN and ULAN, or LCSH - for subject access. A lightweight knowledge-organisation model such as SKOS lets you publish and reuse those vocabularies as linked data.

In practice. Make required metadata a gate, not a suggestion. Configure ingest forms so an asset cannot be published without a title, creator, rights statement and at least one controlled subject term. A short mandatory core beats a long optional schema nobody completes.

Key takeaways

  • Capture metadata and fixity at ingest - it is far cheaper and more reliable than retrofitting.
  • Manage descriptive, technical, administrative/rights and structural metadata, using embedded IPTC/EXIF/XMP where possible.
  • Control subject access with taxonomies and thesauri (Getty AAT/TGN, LCSH) published via SKOS.

Check your understanding

Last modified: Saturday, 1 August 2026, 10:37 AM