Documenting your data
Summary (why and how)
The nature and the extent of the metadata elements will vary depending on the discipline and the chosen metadata standard (see the metadata standards section below). However, the general-purpose Dublin Core Metadata Element Set includes these 15 recommended elements (some may not apply to your dataset):
- Title - the name given to the dataset
- Creator - entity (person, organization or service) primarily responsible for creating the dataset
- Contributor - entity (person, organization or service) who contributed to the creation of the dataset
- Publisher - the entity (person, organization or service) responsible for making the dataset available
- Subject - subject terms or keywords that describe the dataset. The best practice is to use a controlled vocabulary or formal classification scheme
- Description - a brief description, or abstract, of the dataset
- Date – date(s) of creation, publication, or revision of the dataset
- Coverage - describes the spatial and temporal extent of the dataset
- Type - the type of object. For data this would typically be "dataset"
- Format - a description of the format or file type(s) of the dataset
- Identifier - a permanent identifier used to locate and identify the dataset
- Language - the language(s) used within the dataset (if applicable)
- Source - A related resource from which the described resource is derived
- Relation - a relational element describing the relationship of this dataset to other objects, collections, or entities. Examples: other datasets; publications based on the dataset
NOTE: These elements describe the data at the study or dataset level. While these elements give an overview of the nature and content of the data object, they are generally not sufficient to make the data reusable. A complete description of a dataset involves metadata at the variable or data element level. See the Metadata Standards section for more information on this.
Instead of devising your own metadata scheme, you should try to identify an existing and recognized metadata standard. If no discipline-specific standard exist, you may want to start with a general standard like Dublin Core or the DataCite Metadata Schema. Also note that some data repositories like Dryad have developed their own metadata schema.
Disciplinary Metadata Standards
The table below shows some of the better known standards: