datadatums

Datadatums is a proposed term for units of information that describe individual data elements within a dataset. In this usage, a datadatum functions as metadata about a specific data point or field, rather than about the collection as a whole. For example, in a table of temperature readings, a datadatum for a single row might include the measurement timestamp, the sensor identifier, the unit of measurement, the data type, the measurement precision, and a quality flag indicating potential anomalies. Collectively, such per-element descriptors compose what some researchers call per-datum metadata.

Datadatums are distinct from dataset-level metadata (which describes the dataset's origin, scope, and methods) though they overlap with data quality attributes, provenance, and measurement standards. They are related to but not synonymous with metadata fields or data dictionaries; they are specifically attached to individual data elements.

Applications include scientific data management, sensor networks, machine learning datasets, and data curation workflows, where per-record provenance and validation enable reproducibility and traceability.

There is no universal standard for datadatums. In practice, they appear as per-record annotations or embedded fields in data structures (for example, JSON records or column-level metadata). They are often managed via a data dictionary or a schema that defines allowed datadatum attributes such as unit, timestamp, source, and uncertainty.

Because datadatums is not widely adopted, terminology can vary and may be redundant with established metadata practices. When introduced, it should be well defined in project documentation and aligned with existing metadata standards to ensure interoperability.