perdataset

Perdataset is a concept in data management that describes the practice of treating each dataset as a discrete governance unit with its own metadata, provenance, licensing, and access controls. In a perdataset approach, catalogs and repositories attach all relevant information to the dataset record, rather than distributing governance information across an entire collection or project. This enables independent discovery, reuse, and stewardship of each dataset.

The term is used in discussions of reproducibility and data governance and is not a formal standard. It appears as a design pattern in some data platforms and repository schemas, where the emphasis is on dataset-level autonomy within larger data ecosystems.

Core components of a perdataset model include dataset-level metadata (title, description, authors, citation), provenance and lineage (how data was collected and transformed), versioning (snapshots and immutable identifiers), licensing and terms of use, access controls, and quality indicators. Keeping these elements at the dataset level supports clear attribution, traceability, and governance, while allowing different datasets within the same repository to follow distinct policies.

Benefits of the perdataset approach include improved discoverability and reuse, precise permissioning, enhanced reproducibility, and better auditability. It supports modular governance, enabling stewardship to adapt to the needs of individual datasets without reorganizing larger collections.

Implementation considerations involve choosing metadata schemas (such as Dublin Core, DCAT, or PROV), assigning persistent identifiers (DOIs or UUIDs), integrating with data catalogs, and establishing robust versioning strategies. Challenges include standardization across platforms, potential cognitive load for curators, and managing licensing fragmentation.

See also: Data provenance, Metadata standards, Dataset versioning, Data governance, Access control.