Institutional data drifts constantly. Departments merge, faculties are renamed, campuses gain and lose autonomy.
And only the institution knows what actually happened — which makes it the authority on its own structure, and makes any registry that does not ask them systematically wrong.
What the record must capture
- A persistent identifier that survives every rename.
- All name variants — official, abbreviated, translated, historical.
- Hierarchy — department within faculty within institution.
- Location.
- Type and status.
- History of structural change with dates.
Verification sources
- The institution itself, for structure and current names.
- Official registers, for status and legal identity.
- The institution's own publications, for name forms in use.
- External institution registries for cross-linking.
- The institution is authoritative about itself and nothing else is.
Structural change
- Record the type — rename, merger, split, absorption. They are not equivalent.
- Record effective dates.
- Keep the superseded entity resolvable, since historical publications reference it.
- Link predecessors and successors explicitly.
- Never delete an entity that has ever been referenced.
Where errors come from
- Name variants creating duplicate entities.
- Similar names merging distinct institutions — worse, and harder to detect.
- Departments assigned to the wrong parent after restructuring.
- Historical records reassigned to the current structure, destroying the history.
- Affiliated bodies included or excluded inconsistently.
Working with institutions
- Give them a route to review and correct their record.
- Ask them to confirm structure periodically.
- Ask them to publish a preferred affiliation form — this reduces matching problems more than any algorithm.
- Notify them of changes you make.
- Institutions generally want their record correct, since it affects their own figures.
Hierarchy and counting
- State whether counts roll up from department to institution.
- Handle hospitals, institutes and affiliated centres explicitly — these are the most disputed cases.
- Be consistent, or figures will not reconcile.
- Publish the hierarchy so institutions can check it.
One thing worth remembering
Ask institutions to publish a preferred affiliation form and tell their researchers to use it.
It costs the institution one page of guidance. Every correctly written affiliation is a matching problem that never has to be solved — which is worth more than any improvement to the matching algorithm.
Câu hỏi thường gặp
What must an institutional record capture?
A persistent identifier surviving renames, all name variants including historical and translated, hierarchy, location, type and status, and history of structural change with dates.
Who is authoritative about institutional structure?
The institution itself — for structure and current names — with official registers for status and legal identity.
How should structural change be recorded?
Record the type since rename, merger and split are not equivalent, record effective dates, keep superseded entities resolvable, and never delete an entity that has been referenced.
Where do institutional data errors come from?
Name variants creating duplicates, similar names merging distinct institutions, departments assigned to wrong parents after restructuring, and historical records reassigned to current structure.
What is the most effective intervention?
Asking institutions to publish a preferred affiliation form — every correctly written affiliation is a matching problem that never has to be solved.
