Metadata and Descriptions for Digital Collections

What metadata is

Metadata is information about an item, kept in structured fields. For a museum object it might include a title, a maker, a date, materials, dimensions, a place of origin, an identifier and a rights statement. For a photograph of that object it might add who took it, when, with what, and under what license it can be used.

Metadata is distinct from a description, though the two sit together. Metadata is typically short, field-based and meant for sorting, filtering and searching. A description is a piece of readable prose, written for a person, that explains what the item is and why it matters. A digital collection needs both.

Why consistency matters

Search and filtering depend on records agreeing with each other. If one record gives the date as “1890s”, another as “c. 1895” and a third as “late nineteenth century”, a search or a timeline will treat them as unrelated. If one record lists the material as “wood” and another as “Wood, oak”, a filter will split them.

Inconsistency tends to creep in over years, as different people enter records in their own way. Small differences compound into a collection that is hard to search, hard to export and hard to move to a new system. A modest amount of agreed practice early on saves a great deal of cleaning later.

Common trouble spots include:

• Names. The same person spelled several ways, or a maker listed as “Unknown”, “unknown” and “Anonymous”.
• Dates. Mixed formats, approximate dates written inconsistently, or dates stored as plain text that cannot be sorted.
• Places. Modern and historical names mixed without explanation.
• Subjects and materials. Free-form entries where a shared list would do.
• Identifiers. Duplicates, or numbering schemes that changed over time.

Core descriptive fields

There is no single required set, and published metadata standards exist for libraries, archives and museums. They vary in detail, so check which one your sector or software uses. Many records, though, share a recognizable core:

• Identifier (a unique number or code)
• Title or name
• Creator or maker
• Date or date range
• Type or object category
• Materials and technique
• Dimensions
• Place of origin or find
• Subject keywords
• Description
• Rights and credit
• Source or provenance, when known

For a small team, a short list filled in reliably is better than a long one filled in patchily. Start with a minimum that you can complete for every item, and add fields only when there is a clear use for them.

Controlled vocabularies and style rules

A controlled vocabulary is an agreed list of terms. When staff pick from a list, the same thing is always called the same thing. Shared lists for subjects, materials and place names are published for many fields, and some institutions build their own for local needs.

Even without formal lists, a one-page style guide helps. It can say how dates are written, how uncertain dates are expressed, whether names are entered surname first, which words are capitalized, how to record unknowns, and which spelling is used. Write down the decisions and apply them from the start. Revisiting old records is slow, so decide early on.

Plan for uncertainty honestly. A field can say “date unknown” or “attributed to” instead of presenting a guess as fact. Consistent ways of showing doubt are themselves a form of good metadata.

Writing descriptions that help

A public description can serve both visitors and search. Some principles:

• Lead with what the item is. Put the plainest identification first, then context.
• Use the words people would search for, not only specialist terms. If a technical term is needed, explain it briefly.
• Keep it factual. Avoid unverified claims and vague praise. If the history is disputed or uncertain, say so.
• Stay concise. A short paragraph is often enough, with longer notes elsewhere for specialist users.
• Write in plain language. The same habits as in plain language for exhibit text apply.
• Avoid repeating the metadata line for line; use the prose to add meaning.

Some older records contain outdated, biased or offensive wording. Handling this thoughtfully is part of the work, and it connects to collection bias and cultural safety. One approach is to keep the original wording in a labeled field for historical purposes while presenting updated language to the public. Decisions like this ideally involve the communities concerned.

Search, discovery and reuse

Well-structured metadata makes online collection portals work better, since filters and facets can only use fields that are filled in consistently. It also supports search engines and other systems that read the data, and it makes it possible to share records with aggregators or partners. Clean data and clear rights statements make open access publishing far easier.

For images and other media, descriptive information also supports accessibility. A written description of what an image shows can serve as alternative text, and the groundwork for audio description starts with the same observations. Keep these descriptions in separate, labeled fields so they can be used appropriately.

Practical habits

1. Agree the minimum field set and a short style guide.
2. Use drop-down lists or controlled terms wherever possible in the cataloging system.
3. Record who entered or changed a record, and when.
4. Review a small sample regularly for drift.
5. Keep a master copy of the data separate from any exhibit export.
6. Check records before they are used on exhibit screens, because mistakes become visible at once.

Metadata work is rarely finished. A collection record is a living document that improves with new research, so build in time for correction and make it easy for staff to flag issues.