Names are unreliable.
A person can publish under initials, a full name, a pen name, a married name or several transliterations. An organisation can change its name. A city can be known by different names across languages and historical periods. A subject can have an everyday phrase, a technical phrase and an older phrase that later becomes obsolete.
If a library treated every variation as a different identity, the catalogue would fragment.
Authority control is the mechanism that holds those variations together.
This article is part of eduKateSG’s How X Works programme and the How a Library Works series.
The shortest useful answer
Library authority control works by separating an entity from the many labels used to refer to it, then maintaining a governed record of preferred names, variant names, relationships and identifiers so that all relevant catalogue records can point toward the same underlying identity.
In simple terms:
- one person may have many names;
- one subject may have many phrases;
- one organisation may have several historical names;
- authority control tries to keep those routes connected.
Why catalogues need more than text matching
Suppose a catalogue contains books by the same author under three forms of name.
If the system simply matches text, a search for one form may retrieve only part of that author’s work. The collection contains the material, the records exist, and the search engine functions—but discovery still fails because identity has been split.
Authority control solves a different problem from ordinary keyword search. Search asks whether strings match. Authority control asks whether strings refer to the same thing.
The authority record
An authority record stores information about an entity used in catalogue description or subject access.
Depending on the entity and system, the record may contain:
- a preferred form of name;
- variant forms;
- dates;
- occupations or fields of activity;
- associated organisations;
- broader, narrower or related subjects;
- earlier and later organisational names;
- language or script variants;
- identifiers;
- sources that justify the identity decision.
The record is not merely a spelling rule. It is a compact identity model.
Preferred form does not mean “the only correct name”
A library chooses a preferred form so records can be grouped consistently.
That does not erase other names.
Variant forms remain valuable because users may search them. A person may be best known publicly by one name while older publications use another. A translated name may be common in one language community while the original script is common in another.
The preferred form gives the system a stable centre. Variants give users multiple routes into that centre.
Disambiguation: when two people share one name
The opposite problem is just as serious.
Two different people may have exactly the same name. If the catalogue merges them, works by one person appear under the other.
Authority control uses distinguishing information—such as dates, field of activity, affiliation or another identifier—to keep identities separate.
This is identity resolution: determining whether two records belong together or must remain distinct.
One entity, many scripts
Global libraries face multilingual identity problems.
A name written in Chinese, Arabic, Cyrillic or another script may have several romanisations. Historical systems may use older transliteration conventions. Publishers may spell the same name differently.
Authority control can connect these variants so users do not need to know the exact form selected by one catalogue.
This makes multilingual discovery a relationship problem rather than a spelling contest.
Corporate bodies change through time
Organisations are especially difficult because identity can change gradually.
A ministry can be renamed. Two institutions can merge. A department can become an independent agency. A university faculty can reorganise. A company can change legal name while maintaining continuity of operations.
Authority records can preserve earlier and later names and record relationships among them.
This matters for historical research. A user researching one modern organisation may need publications created under its former name.
Subjects need authority control too
Subject language changes.
Everyday terminology may differ from scholarly terminology. Terms can become outdated, too broad or socially inappropriate. New concepts appear before established vocabularies catch up.
A controlled subject system selects authorised terms and records relationships to alternatives.
A user can search a non-preferred phrase and still be directed toward material indexed under the preferred subject heading.
This is one reason classification and authority control are closely connected. Both create navigable structure, but at different layers.
Broader, narrower and related terms
Subject authority records can do more than standardise labels. They can model conceptual relationships.
- A broad term can lead to narrower specialisations.
- A narrow term can lead back to its broader context.
- A related term can lead sideways into a neighbouring concept.
This turns the vocabulary into a small knowledge graph.
The user is no longer forced to guess one exact phrase. The system can suggest routes through the conceptual neighbourhood.
Authority control and the catalogue
Catalogue records describe books and other resources. Authority records describe the entities those catalogue records point to.
Many catalogue records may therefore refer to one authority record.
When an author’s preferred form changes, the authority system can provide a controlled way to update or reinterpret many associated records without treating every bibliographic record as an isolated case.
This creates centralised identity governance.
Why identifiers matter
Names remain useful for humans, but identifiers are often stronger anchors for machines.
A stable identifier can distinguish one person from another even if their names are identical. It can connect records across languages and systems even when the displayed label differs.
Identifiers also help authority records link outward to other trusted systems.
This lets a local library benefit from a larger identity network rather than maintaining every fact alone.
Shared authority files
Libraries cooperate because the same author, organisation or subject appears in collections around the world.
Shared authority services allow institutions to reuse established identities and contribute improvements.
This reduces duplication of effort and improves interoperability.
It also requires governance. A shared error can propagate widely, so evidence and correction processes matter.
Authority control is evidence-based
Identity decisions should not be arbitrary.
Cataloguers consult the resource itself, publisher information, reference sources, institutional records and other trusted evidence.
When uncertainty remains, the authority record can preserve notes or distinctions rather than pretending the ambiguity has disappeared.
Authority control therefore contains a miniature research process: claim, evidence, decision and revision.
What happens when the preferred name changes?
Authority systems must change without breaking history.
A person may request a different form of name. An organisation may officially rename itself. A subject heading may be revised because terminology has evolved.
A strong system changes the current preferred label while preserving cross-references from earlier forms.
This protects continuity. Old citations and searches still have somewhere to go.
Historical names are data, not clutter
Older terminology may be unacceptable or obsolete today and still be important for historical retrieval.
If a historical archive used an earlier institutional or subject name, researchers may need that term to locate period material.
The library therefore distinguishes endorsement from recordkeeping. Preserving an old term as a variant does not require treating it as the preferred modern label.
Authority control and bias
Controlled vocabularies reflect institutional history.
Some terms and hierarchies were created under assumptions later recognised as narrow, exclusionary or inaccurate.
Updating authority language can improve both respect and retrieval, but changes must be documented so historical records remain traceable.
The authority file is therefore not a frozen dictionary. It is governed infrastructure that must learn.
Authority control in digital libraries
Digital libraries amplify the value of authority control because their collections can be searched and connected at enormous scale.
If identity is clean, a user can move from one author to all works, from one organisation to its archival records, from one place to maps and photographs, or from one concept to resources across several repositories.
If identity is weak, digital scale produces digital confusion faster.
Linked data changes the shape of authority control
Traditional authority systems often store a preferred heading plus cross-references.
Linked-data systems can represent the authority entity itself as a node with relationships to other entities.
- a person is affiliated with an organisation;
- a person created a work;
- an organisation succeeded another organisation;
- a place is located within another place;
- a subject is narrower than another subject.
This makes authority control increasingly central to knowledge-graph architecture.
Authority control and search ranking
Authority information can improve search beyond exact matching.
A query using a variant name can be expanded toward the preferred identity. Records under several forms can be grouped. Ambiguous names can be separated into distinct entity choices.
This produces a better user experience because the search interface can ask a meaningful question: “Which John Smith do you mean?” rather than returning one undifferentiated list.
Authority control and citation discovery
Research citations are messy.
Initials may replace full names. Diacritics may disappear. Journal databases may abbreviate names differently. Older records may use earlier institutional affiliations.
Authority control helps collapse these variants toward a coherent research identity, improving both discovery and bibliometric analysis.
Machine-generated identities
Modern systems can use algorithms to suggest that two records refer to the same entity.
They can compare names, dates, affiliations, co-authors, subjects and identifiers.
This can accelerate authority work, especially in large digital collections.
But automated merging is risky. Combining two different people can contaminate many records at once. Strong systems therefore preserve confidence, evidence and human review for ambiguous cases.
Authority control and AI
AI makes authority control both easier and more necessary.
Language models can recognise aliases, transliterations, historical names and contextual clues that simple exact matching misses.
At the same time, generative systems can confidently merge entities that merely look similar.
A governed authority layer gives AI something stable to resolve against.
This is one of the strongest future roles for library infrastructure: not just helping humans find books, but helping machines know which person, organisation, place, work or concept they are actually talking about.
Authority control is a memory system
At first glance, authority control looks like tidying names.
At a deeper level, it preserves continuity through change.
A person changes name. An organisation changes structure. A place changes jurisdiction. A subject changes terminology. The authority system remembers enough of the previous state that users can still traverse from old records to new understanding.
It is therefore temporal infrastructure for identity.
The complete mechanism
- A catalogue or repository encounters a person, organisation, work, place or subject that requires controlled identity.
- Evidence is gathered from the resource and trusted external sources.
- The system determines whether the entity already exists.
- If it does, the new record links to the existing identity.
- If it does not, a new authority record is created.
- A preferred label is selected according to policy.
- Variant names and historical forms are retained as access routes.
- Identifiers and relationships strengthen disambiguation.
- Catalogue records point toward the authority entity instead of relying only on raw text.
- Search systems use variants and relationships to improve retrieval.
- Changes in names, terminology or knowledge update the authority layer without discarding historical continuity.
- Shared authority networks allow many libraries and digital systems to reuse the same identity infrastructure.
That is how library authority control works.
It is the quiet machinery that prevents a collection from forgetting that different names can belong to the same thing—and that the same name can belong to completely different things.