CurioAI
Manuscripts and archives

Manuscript cataloging: describe the object and its intellectual context

Manuscript cataloging must describe both intellectual content and the physical object. A useful record connects title or supplied title, creator, dates, language, extent, arrangement, provenance, access conditions, and evidence visible in the manuscript itself.

01

Identify the level and purpose of description

Decide whether the record describes one codex, one document, a file, a series, or a whole collection. Archival description preserves relationships between levels; item-level description should not erase the context supplied by a collection or finding aid.

Create a stable identifier or shelfmark and record the repository, collection, and parent description. This keeps images and extracted text connected to their intellectual context.

02

Describe content without neglecting material evidence

Record titles, creators, dates, language, script, subjects, names, and a scope note appropriate to the chosen level. Also document support, extent, dimensions, binding, foliation or pagination, decoration, watermarks, seals, annotations, and condition when relevant.

Dates and attributions may be uncertain. Preserve supplied wording, distinguish inferred dates, and name the source of an identification.

03

Photograph views that support verification

Capture covers, spine, representative openings, colophon, ownership marks, decoration, watermarks where possible, damage, and a scale view. File labels should connect every image to the shelfmark and folio or page.

For live capture, a manuscript-specific extraction plan can ask for missing views and fields rather than treating the object like a coin, ceramic, or painting.

04

Map to the receiving system before export

DACS, ISAD(G), EAD, MARC, MODS, and local systems address different descriptive and exchange needs. Do not claim standards compliance from field names alone; document the mapping, required elements, hierarchy, authority control, and validation rules.

Use a reviewed crosswalk and test a small export. CurioAI can provide structured JSON, CSV, Dublin Core XML, and LIDO XML as starting points, but institutional ingest profiles still require professional mapping.

Before you finish

Review checklist

  • The level of description is explicit.
  • Shelfmark and repository context are stable.
  • Content and physical structure are both described.
  • Supplied and inferred dates are distinguished.
  • Names and subjects use documented authorities.
  • Images identify folio, page, and view.
  • Export mappings are validated for the destination.

Where AI fits

Technology can accelerate capture, transcription, field suggestions, and consistency checks. It cannot establish provenance, legal title, cultural authority, or conservation conclusions from an image alone. CurioAI keeps generated fields editable and requires a person to review the record before saving it.

See the guided cataloging workflow
Further reading

Sources and professional guidance

These sources inform the workflow above. Institutions should also follow their own documentation policy, legal requirements, community protocols, and specialist standards.

From guidance to practice

Create a schema for the collection in front of you.

Use photographs, live questions, and an editable extraction plan to build a reviewable catalog record.

Request access