Metadata enrichment

Metadata enrichment is the process of adding descriptive, administrative, or relational metadata elements to a research record after its initial deposit or harvest, typically by resolving names and free-text values against authority files and controlled vocabularies (ORCID, ROR, the Crossref Funder Registry, subject-classification schemes) or by pulling additional attributes from an external API (citation counts, open-access status, computed topic classifications). It is the step that turns a minimally complete core metadata record — enough to be found and cited — into one that is genuinely interoperable and linkable across systems.

How it differs from data cleaning and initial deposit

Enrichment is easy to conflate with two adjacent but distinct activities. Initial deposit is the act of creating the baseline record in the first place — an author submitting a manuscript, a repository harvesting a DOI record from Crossref. Nothing is enriched yet; the record simply exists. Data cleaning — normalizing a misspelled author name, standardizing a date format, deduplicating two records for the same output — corrects or reconciles values already present but adds no new elements. Enrichment specifically adds new, previously absent metadata — a resolved ROR ID for an affiliation string, an ORCID iD for an author name, a subject classification, a funder identifier — usually sourced from a separate, authoritative system rather than typed in by the depositor.

Worked examples

Example 1 — CRIS ingest pipeline. A CRIS harvests a publication record from Crossref containing only a DOI, a free-text author-name list, and free-text institutional affiliations. The CRIS’s ingest pipeline then enriches that record by querying the ORCID API to resolve each author name to a disambiguated ORCID iD, and querying the ROR API to resolve each affiliation string to a ROR ID — writing both identifiers back into the record as new fields that did not exist at deposit. This is the same identifier-resolution logic described in CASRAI’s identifier crosswalk entry, applied specifically as an enrichment step in an ingest workflow.

Example 2 — OpenAlex. OpenAlex ingests raw bibliographic metadata originally deposited with Crossref or PubMed, then enriches each Work record with elements the original deposit never contained: a computed topic/subject classification, a citation count aggregated from OpenAlex’s own citation graph, and an open-access status resolved via Unpaywall. None of these fields exist in the Crossref-deposited record itself — they are added downstream, which is precisely what makes this enrichment rather than harvesting. See CASRAI’s OpenAlex guide for more on how that pipeline works.

Counter-example

Correcting “Jon Smith” to “John Smith” in an existing author field is data cleaning, not enrichment — the field already existed and held a value; only its accuracy changed, and nothing new was added to the record. Likewise, the original act of depositing core metadata (title, creator, date, identifier) at time of publication is not enrichment — it is the baseline record enrichment subsequently builds on.

Why it matters for CRIS and repository interoperability

Enrichment is what makes cross-system linking actually work in practice. A record with only free-text author and affiliation strings cannot be reliably joined to an author’s other outputs, a funder’s grant portfolio, or an institution’s research output count — string matching on names is fundamentally ambiguous. Once a record carries resolved PID metadata (ORCID iDs, ROR IDs, DOIs), it becomes machine-joinable across systems without manual reconciliation, which is the practical basis for CRIS interoperability. This is also why enrichment is typically an ongoing, repeatable pipeline step rather than a one-time task — source authority files (ROR’s organization data, ORCID records, funder registries) are updated continuously, so records enriched once can be re-enriched as those upstream sources improve.

References

• OpenAlex, “About” and Works documentation (openalex.org, help.openalex.org).
• ROR, “ROR Data Schema and Crosswalks” (ror.org).
• ORCID, API documentation on record matching and affiliation data (info.orcid.org).