Authority Control at Scale: One Person, One Authority, Across Every Collection

A researcher searches your catalogue for a local figure — say, a mill owner who turns up across your business records, a photograph collection, a council minute book, and a bundle of family correspondence. In a well-run archive, that one search returns everything. In most archives, it does not. The mill owner is catalogued as “Smith, John” in one collection, “John Smith (1841–1912)” in another, “J. Smith, Esq.” in a third, and “Smith, J., manufacturer” in the fourth. To your database, those are four different people. To the researcher, the trail has gone cold.

This is the problem authority control exists to solve, and it is one of the least glamorous but most consequential parts of archival description. Authority control is the discipline of deciding that a person, organisation, or place has one agreed, controlled form of name — one authority record — and that every mention of them across every collection points to it. Get it right and your holdings become genuinely searchable. Get it wrong, or skip it, and even a fully catalogued archive hides its own connections.

Why inconsistent names break discovery

The damage from inconsistent name forms is quiet but severe. Names vary for entirely innocent reasons: married and maiden names, anglicised and original spellings, titles and honorifics, initials versus full forenames, organisations that merged or renamed. Two cataloguers working a year apart, each making reasonable choices, will produce different forms for the same person. Neither is wrong in isolation. Together they fragment the record.

The consequence is that cross-collection discovery — the thing that makes an archive more than the sum of its boxes — simply fails. The relationships that give a collection its research value are precisely the ones that span collections, and those are exactly the relationships inconsistent naming hides. The most interesting connection in your holdings might be the one no single search can surface, because the person sits under four different headings.

This compounds the hidden cost of uncatalogued and inconsistently catalogued collections. It is not enough to have described the material; the descriptions have to agree with one another about who is who.

What reconciliation against VIAF, FAST, and Wikidata gives you

The solution is not to invent your own private list of correct names and hope it stays consistent. It is to reconcile your people, organisations, and places against established, shared authority files — and to do so as you catalogue, not as a someday-cleanup project.

Three of these matter most for archives:

• VIAF, the Virtual International Authority File, aggregates the name authority files of national libraries worldwide. Matching a person to their VIAF identifier ties your record to a globally recognised form of their name.
• FAST (Faceted Application of Subject Terminology), derived from the Library of Congress Subject Headings, provides controlled forms for subjects, places, and named entities in a structure designed to be machine-applied.
• Wikidata offers a richly linked, openly licensed identifier for an enormous range of people, organisations, and places, and acts as a hub connecting to dozens of other authority systems.

Reconciliation means matching the “John Smith” in your record to the specific, identified John Smith in these files — and storing that identifier alongside the name. Once you do, you have a linked authority, not just a string of text. The name is now anchored to a stable identifier that other systems recognise.

Archivers.ai performs this reconciliation as part of the description workflow. As it proposes the people, organisations, and places named in an item, it can match them against VIAF, FAST, and Wikidata, so that the authorities are built in from the first record rather than retrofitted across thousands. Because the platform is the processing and access layer in front of your existing catalogue — AtoM, ArchivesSpace, CALM, Axiell — those reconciled authorities flow into the system you already run, rather than living in yet another place.

Linked authorities are what make discovery work

The shift from name-as-text to name-as-linked-authority is the whole point, and it changes what is possible.

When every mention of one person resolves to one authority, a single query genuinely returns everything about that person across your holdings — the business records, the photographs, the minutes, the correspondence — regardless of how each was originally worded. That is cross-collection discovery working as it should.

Linked authorities also let your catalogue connect outward. Because the identifiers are shared, your records can participate in the wider web of cultural heritage data — aggregators and discovery services that key on VIAF or Wikidata identifiers can surface your material alongside related holdings elsewhere. An archive with reconciled authorities is not an island; it is part of a network. And because the underlying identifiers are stable, the connection survives even as display name forms change over time.

Keeping the archivist in control

Reconciliation is a place where automation can go wrong if it is left unsupervised, and the failure mode is specific: matching the wrong John Smith. Authority matching is a judgement, especially for common names and for figures who are poorly represented in the reference files. A confident-looking but incorrect match is worse than no match, because it asserts a connection that does not exist.

This is why authority control belongs inside a review gate. Archivers.ai proposes a reconciliation; the archivist confirms or rejects it. Every suggested match carries a per-field confidence flag, so a cataloguer’s attention goes straight to the uncertain cases — the common surnames, the ambiguous initials — rather than being spread evenly across matches that are obviously right. AI proposes; the archivist disposes. The provenance of each decision is recorded, so the authority work is defensible later, not a black box.

When the authorities are settled, they export cleanly. Archivers.ai produces standards-aligned records — EAD3, Dublin Core, ISAD(G)-structured descriptions — with the reconciled identifiers carried through, so your authority work is portable and not stranded in a single tool.

Worth doing, and worth doing as you go

For professional archivists and record offices, authority control at scale has long been the right thing that there was never quite time to do. Reconciling against VIAF, FAST, and Wikidata as part of cataloguing — rather than as a separate, perpetually deferred project — finally makes it tractable. For community archives, where the people, places, and organisations in the collection are often the entire point, linked authorities are what turn a set of records into a resource the whole community can actually navigate.

One person, one authority, across every collection. That is not a tidiness preference. It is the precondition for an archive that answers the questions people bring to it.

If you want to see reconciliation in your own material, join the Archivers.ai waitlist for early access on a sample collection — and test authority control across a larger body of records when you are ready.