7AAVDM32 Chap.8 Taxonomies and Controlled Vocabularies
Taxonomies and Controlled Vocabularies
Finding paths with defined relations
This chapter constructs taxonomy facets from user tasks and tests how one asset can sit under audience, format and subject routes without becoming duplicate objects. It distinguishes synonyms from broader, narrower and related concepts.
Begin with user tasks
A taxonomy should help named users make distinctions that matter to their work.
Interview, search-log and task evidence can reveal whether people need to browse by subject, audience, asset type, campaign, region or another facet. Do not begin by copying an organisational chart or turning every metadata value into a hierarchy.
Define the retrieval decision first, then choose a structure that supports it.
Keep organising principles coherent
Siblings in a hierarchy should usually answer the same kind of question. Mixing audiences, formats and topics beneath one parent produces ambiguous paths and makes coverage difficult to test. Facets let these dimensions remain separate while combining during retrieval.
An asset can then be described once and reached through several routes without creating competing duplicate records.
Relationships have distinct meanings
Synonyms connect different labels for the same concept; broader and narrower relations express hierarchy; related relations connect associated concepts without claiming parentage. Conflating them can expand or suppress search incorrectly.
Define each preferred term, record permitted alternatives, and test whether a user can explain the relationship in plain language. A relationship that cannot be justified should not be added merely because two terms often occur together.
Boundary cases reveal design quality
Easy examples rarely expose overlap, missing coverage or inconsistent granularity.
Test items that plausibly fit several categories, sit between levels, or use contested language. Record why the chosen placement is appropriate and whether multiple assignment is legitimate. These examples become governance evidence and help editors apply the structure consistently when new assets arrive.
Vocabulary control is continuing work
Terms change as collections, users and organisational purposes change.
A maintained vocabulary needs an owner, proposal route, definition standard, approval authority, version record and migration plan for affected assets. Search mappings can preserve discovery when a preferred term changes.
Review should examine both technical consistency and whether the structure still represents communities fairly.
Govern language and consequence
A bias audit traces how preferred labels, omissions and hierarchical placement shape what users can find and how subjects are framed. Ask whose terminology has authority, who is absent, what a change would do to existing records, and how a person can challenge a term.
The final build exercise adds definitions, authority, mappings and change control. Its success criterion is not a tidy tree alone, but a governed structure that supports retrieval and can be revised responsibly.
What this chapter covers
- 01
Taxonomy
- 02
Controlled vocabulary
- 03
Polyhierarchy
- 04
Scope notes
- 05
Preferred terms
- 06
Vocabulary governance
Worked example · free
Repair an audience taxonomy
- 2Identify the mixed organising principles.
- 2Separate audience and knowledge facets.
- 3Define mappings and review rules.
Key terms
- Taxonomy
- A Taxonomy organises concepts into defined categories and relationships for navigation and classification.
- Controlled Vocabulary
- A Controlled Vocabulary manages preferred terms, synonyms and relationships to support consistent description.
- Polyhierarchy
- Polyhierarchy allows one concept or asset to appear through more than one legitimate hierarchical route.
Taxonomies and Controlled Vocabularies FAQ
Why are folders not a complete taxonomy?
Folders often force one location and hide alternative discovery routes. Metadata relationships can connect one authoritative asset to several subjects, audiences or formats. The benefit depends on clear rules; uncontrolled multiple placement can create noise and maintenance burden.
What makes taxonomy categories usable?
Define the organising principle, scope, preferred label, boundaries and examples, then test representative and awkward assets. Users and cataloguers should apply the categories consistently enough for the intended task, even though no classification captures every perspective.
How should a legacy harmful term be handled?
Document the challenge, consult relevant authority, choose a current preferred display or contextual treatment, retain the old value only where audit and retrieval require it, and migrate affected records consistently. Publish a route for future review.
How can I tell whether a hierarchy needs separate facets?
Check whether sibling categories answer the same kind of question. If one branch describes audience, another format and another subject, users cannot predict placement and cataloguers will apply inconsistent logic. Separate the dimensions into facets, define each value and test combinations on boundary assets. Keep one authoritative asset record while allowing legitimate values from several facets.
The result should improve retrieval without manufacturing parent-child relationships between concepts that are merely associated.
Assessment move
Build a tiny taxonomy from one user task, then challenge it with an asset that fits two branches and a term that affected users contest. Document whether each repair needs a synonym, relationship, separate facet or governance decision.