What Do We Mean by ‘Version’? Actionable Guidelines for the Versioning of Data

The data lifecycle, from acquisition to release, is increasingly complex, involving multiple processing stages, research groups, and funding sources. There are also growing concerns about data sovereignty and data governance. This complexity demands a robust conceptual framework to ensure reproducible versioning and linking of any original datasets to their many derivatives.

To address this, we initially developed six data versioning principles for digital data artefacts by analysing use cases and adapting the Functional Requirements for Bibliographic Records framework for analogue Information Resources. Our six principles established a common language for key concepts and terms: Revision, Release, Granularity, Manifestation, Provenance, and Citation.

In this paper, we present recommendations for implementing the data versioning principles after consulting data communities. The recommendations cover: adopting a consistent versioning strategy; considering standardisation initiatives; using persistent identifiers and unambiguous identification of versions; implementing clear and descriptive version labels; ensuring user-friendly version control systems; documenting changes and metadata; and communicating versioning practices clearly to stakeholders.