2.1 Metadata
2.1.1 Definition
Metadata refers to information that describes, explains, locates, or otherwise makes it easier to retrieve, interpret, use, or manage data.[1] It may also include information that describes how data should be published, managed, and displayed. Metadata must be made accessible in forms and formats that are convenient for both users and machines.
It is essential that metadata accurately matches the dataset it represents and gets updated anytime the data changes. This makes keeping versions of data and metadata files important, with semantic versioning being the suggested approach.
2.1.2 Types
Metadata encompasses several categories, each fulfilling a distinct role. Certain types of metadata facilitate data users in locating, evaluating, interpreting, and utilizing datasets efficiently, whereas others are intended to assist administrators of data repositories in the management and distribution of data.
Metadata relevant to data users
Descriptive metadata identifies and describes a dataset, typically including elements such as the dataset title, abstract or summary, relevant keywords and topics, information on producers and publishers, geographic and temporal coverage, and versioning details. Descriptive metadata encompasses cataloguing metadata, a subset used specifically to register and manage datasets within catalogs or repositories, which is vital for ensuring datasets are both findable and citable.
Reference metadata offers contextual and methodological details about the dataset. Common components include conceptual definitions, sources of data, methodologies employed, sampling and processing procedures, quality assessments, and known limitations.
Structural metadata specifies the format and organization of the dataset, as well as relationships among its elements, facilitating effective analysis and integration into workflows. For microdata, structural metadata may consist of a data dictionary detailing files and variables; for indicators, it involves a data structure definition describing the data model and structure; and for geographic datasets, it takes the form of a feature catalog.
Collectively, descriptive, reference, and structural metadata empower data users to evaluate a dataset’s suitability for their needs, comprehend its scope and coverage, and utilize it appropriately. These metadata types are indispensable for data producers and curators, as they underpin the usability, reliability, and discoverability of datasets. The creation and maintenance of such metadata is mandated by the Development Data Quality Policy and constitutes a primary focus of the standards detailed in this guide.
Metadata intended for catalog and systems administrators
Metadata relevant to catalog and systems administrators is tailored to the specific IT environments in which catalogs and systems function. As such, no universal standard applies to these metadata types; however, it remains essential that they are structured and formatted consistently alongside descriptive, reference, and structural metadata.
Administrative metadata encompasses information supporting dataset management, security, and governance by catalog and system administrators. This includes details regarding dataset origin, acquisition date, and access rights, such as licensing terms and permissions. Administrative metadata defines conditions for publishing or accessing datasets, specifying authorized users, permitted purposes, and applicable constraints. Additionally, it captures operational and technical requirements necessary for effective dataset administration—including file formats, storage needs, and system dependencies. Proper documentation of administrative metadata ensures secure, efficient, and compliant management in accordance with institutional policies and procedures. Its context-specific nature means the structure and content will vary among organizations, and no standard is mandated.
Presentation and portrayal metadata establish protocols for the visual representation of data, promoting clarity, consistency, and accurate interpretation within dissemination platforms and data catalogs. As optional elements—since many systems default to preset presentation and portrayal parameters—presentation metadata (sometimes referred to as display or visualization metadata) describes general rules for non-spatial data displays, such as chart types, color schemes, symbols, axis labels, legends, and formatting conventions. These components facilitate adherence to organizational standards and visualization style guides and consistent rendering across platforms. Portrayal metadata, conversely, focuses on the visualization of geographic or spatial datasets by providing specific cartographic instructions, including symbology, feature styling, color palettes, scale-dependent rendering, and related elements. Collectively, presentation and portrayal metadata ensure statistical and spatial data are presented in a clear, standardized, and comprehensible manner.
Usage and impact metadata serve distinct functions in understanding dataset engagement and influence. Usage metadata tracks how datasets are accessed and utilized, gathering metrics such as download counts, API usage, page views, and geographic patterns of user activity. Impact metadata documents instances where datasets contribute to research, policy formation, or decision-making, reflected in citations, references, or indicators of user engagement and reach. Combined, these metadata dimensions offer valuable insights into dataset relevance and effectiveness. For institutions such as the Bank, this supports trend analysis, user experience optimization, and enhances visibility and credibility in the global data environment. They further contribute to catalog ranking and resource discovery strategies. Usage data should be collected via appropriately configured web analytics tools for real-time reporting, and while integration into records is optional, automated dynamic updates are recommended for accuracy. Impact metadata generally requires ongoing monitoring of external dissemination channels, such as publication databases.
Relationship metadata articulate links between datasets and associated resources, playing a critical role in documenting data lineage and dependency. This metadata identifies source datasets, derivative outputs, and connections to supplementary materials (e.g., publications, methodological documents, code repositories). For example, relationship metadata may associate a working paper with its referenced original dataset, derived analytical datasets, and reproducible scripts used in processing. Maintaining these connections enables catalog administrators and repository managers to develop integrated, interoperable systems that foster discoverability, traceability, and data reuse. Relationship metadata facilitating data interpretation and supporting reproducibility should be made accessible to end users.
Definition adapted from Understanding Metadata, National Information Standards Organization, 2004. ↩︎