Why provenance matters in scholarly metadata

A publication title, citation count or affiliation may look simple on screen, yet it can come from several different providers. When those providers disagree, a platform needs to know which source supplied each piece of evidence and when it was last refreshed.

Provenance turns metadata from a static claim into an inspectable evidence trail. It helps maintainers compare conflicting values and helps readers understand that some fields are externally sourced rather than authored by ResearchHub.

Not all matches deserve the same confidence. A DOI match is generally stronger than a loose title similarity. A verified institutional identifier is stronger than a free-text affiliation string. ResearchHub uses these differences to decide which relationships should be public and which should remain candidates for review.

Confidence is especially important when connecting people and institutions, where a false positive can misattribute work or affiliation.

Scholarly metadata changes over time. Articles receive issue assignments, open-access status can change, affiliations can be clarified and citation information can be refreshed. If a platform only stores the latest visible value, it becomes harder to diagnose why a record changed.

Keeping provider and synchronisation information allows maintainers to distinguish a legitimate update from a faulty import and makes targeted repair safer.

No large scholarly graph will be free of errors. Trust therefore depends not only on preventing mistakes but also on making them correctable. Provenance gives users and maintainers a common starting point for a correction: which source said what, and which evidence should replace it?

A platform that exposes provenance can improve continuously without pretending that imported metadata is infallible.