Check Schema Compatibility Before Combining Repository Exports

Identify the contract behind each export

Save a small raw sample from every repository and record the metadata schema version declared in the export. A parser built around an older list of resource types can silently drop newer values. Read the release notes for the relevant versions and make a mapping table with the original field, allowed value, local destination, and handling rule.

DataCite's 4.6 release documents additions including Award and Project resource types and a translation relationship pair. Use that historical version as a concrete test case when evaluating an import that claims to accept 4.6. Preserve unfamiliar values in a review field so that future schema changes do not quietly reduce the number of usable records.

Test a hypothetical import

An export contains 100 records, including seven projects and three awards. An older importer recognizes the other 90 and discards unfamiliar types. If the final dataset has 90 rows, the apparent 10% loss comes entirely from type handling. Count both rejected records and accepted records by raw type before considering any change to the scientific inclusion criteria.

Create a small fixture containing each affected type and compare the imported rows with the original export. Retain version-specific mappings and test a known older record as well. A converter that preserves both raw values and normalized categories makes later reclassification possible without repeating the retrieval.

DataCite Metadata Schema 4.6 release notes

Related worksheet: Table grain and primary-key worksheet.

Source checked on 5 November 2026.