Controlled Vocabulary

Why free tagging breaks down

When each person types their own tags, one concept ends up with several names. A typical marketing library contains variants like these:

Concept	Variants found in free-text tags
Product photo on a plain background	packshot, pack shot, product shot, product image, PS
Germany as a market	Germany, DE, GER, Deutschland
Trade show material	trade show, tradeshow, trade fair, expo
Sales presentation	sales deck, pitch deck, presentation, slides

A filter on any one variant misses the assets tagged with the others, and nobody can tell how many were missed. A controlled vocabulary removes the variation at the point of entry. It also makes metadata reliable enough for software to act on: automation rules, access rules, partner portals and integrations can depend on a term that is always spelt and scoped the same way.

Types of controlled vocabulary

Type	Structure	Example
Term list	A flat list of allowed values	Asset type: packshot, lifestyle photo, banner, sales deck
Synonym ring	Terms treated as equivalent in search	sofa, couch, settee
Taxonomy	Terms arranged from broader to narrower	Furniture > Seating > Office chairs
Thesaurus	A hierarchy plus preferred terms, synonyms and related terms	"Sofa" preferred over "couch", related to "armchair"

ANSI/NISO Z39.19-2005 gives guidelines for building monolingual controlled vocabularies of all four kinds, and ISO 25964-1 (2011) covers thesauri. On the web, W3C's SKOS (2009) is the usual format for publishing them. Most DAM vocabularies are term lists and taxonomies; full thesauri are more common in libraries and publishing.

What to put under vocabulary control

Controlled vocabularies suit the parts of a library that are finite and change slowly:

• asset type, channel and format;
• product line and product, brand or sub-brand;
• region, market and language;
• campaign and season;
• licence type and approval status.

Captions, descriptions and the long tail of subjects ("woman at a desk", "autumn leaves") stay as free text. Auto-tagging and semantic search cover that part better than any list can.

Maintaining a controlled vocabulary

• Name an owner. A small group adds and changes terms, and everyone else requests new ones. A vocabulary anyone can extend turns back into free tagging; one that nobody maintains falls out of date.
• Give each term a stable code separate from its label. Labels can then be corrected or translated without breaking filters, saved searches or integrations.
• Translate labels, not terms. One term with a label in each language means a French and a German colleague tag the same term, and a filter in either language finds both.
• Decide what happens to assets when a term is retired. Move them to a replacement term or clear the value. Deleting the term alone leaves values nobody can filter on.
• Review usage once or twice a year. Terms that nobody applies or searches for can usually be merged into another term or removed.

Controlled vocabularies in Freedam

• Administrators create vocabularies as flat lists or as trees, and a flat list can be converted into a tree later. See taxonomy.
• Each term has a stable code and a label per language. Missing labels can be translated automatically into the workspace's languages.
• A vocabulary is attached to metadata fields as a single or multiple choice, and one field can depend on another, so picking a product line limits the product field to the products in that line.
• Deleting a term that assets still use asks what to do with them: move them to a replacement term or clear the value.
• A vocabulary can be marked as protected, so that changing or deleting it needs an extra confirmation.
• Vocabulary fields can be shown as quick filters next to the search bar in the asset library.
• Analytics insights list terms that, 60 days after they were created, have never been applied to an asset, searched for or used as a filter.

Frequently asked questions

What is the difference between a controlled vocabulary and a taxonomy?

A controlled vocabulary is any agreed list of terms. A taxonomy is a controlled vocabulary whose terms are arranged in a hierarchy from broader to narrower. Every taxonomy is a controlled vocabulary; a flat list of asset types is a controlled vocabulary but not a taxonomy.

What is an example of a controlled vocabulary?

A list of asset types (packshot, lifestyle photo, banner, sales deck, datasheet) that people must pick from when they upload a file. Country lists based on ISO 3166 codes and language lists based on language codes are other common examples.

Can AI tags replace a controlled vocabulary?

No. AI tags describe what an image shows, such as "chair" or "office". They cannot know your product names, campaigns, markets or licence terms. Use AI tags for visual subjects and controlled vocabularies for the business facts people filter on.