Metadata is what makes a digitised item findable, understandable, manageable and preservable. Without it a perfect scan is an orphan file. A digitisation project must plan its metadata as deliberately as its imaging - deciding which schemas apply, who supplies each element, and how it is captured without hand-keying the same value twice.

International practice recognises several distinct kinds of metadata, each doing a different job across the object's life. Confusing them - or capturing one and neglecting the others - is a common and expensive mistake.

The types of metadata

  • Descriptive - who, what, when, where; supports discovery. Dublin Core for interoperability, EAD for archival finding aids, MODS for library-rich description, Spectrum for museum collections.
  • Structural - how parts relate (page order, volumes, multi-file objects), often expressed with METS.
  • Administrative & technical - capture device, resolution, colour profile, file format and rights.
  • Preservation - fixity, provenance and format events, expressed with PREMIS.

Standards and identifiers

Use controlled vocabularies and authority files (e.g. LCSH, AAT, VIAF, geonames) so names and subjects are consistent and linkable. Assign persistent identifiers - such as ARKs, DOIs or Handles - so items can be cited and resolved even when systems change. Consistency here is what enables aggregation and linked-data reuse later.

In practice. Capture technical metadata automatically at the point of scanning (embedded EXIF/XMP, tool-generated PREMIS events) and reconcile descriptive metadata against your existing catalogue. Every field a human types by hand is a field that will contain errors and slow throughput.

Indexing for discovery

Indexing exposes metadata and full text to search. Map your descriptive fields to the search index, apply the controlled vocabularies as facets, and combine OCR/HTR text with structured fields so users can both browse and keyword-search. Well-modelled metadata is what lets a search engine rank and cluster results meaningfully.

Key takeaways

  • Plan descriptive, structural, technical and preservation (PREMIS) metadata together.
  • Use recognised schemas (Dublin Core, EAD, MODS, METS, Spectrum) and authority files.
  • Automate technical capture and assign persistent identifiers for lasting citability.

Check your understanding

Last modified: Saturday, 1 August 2026, 9:38 AM