Beginner Glossary#
This glossary gives short meanings in the context of OKF Explorer. Follow the linked chapters for the important qualifications.
For the curriculum's stable, detailed definitions, use:
- assertion, authority and channel;
- class, confidence, controlled vocabulary and data;
- DIKW, evidence, identifier, information, instance and IRI;
- jurisdiction, knowledge, licence, life event and observation time;
- ontology, outcome, predicate, property, provenance and public service;
- redress, requirement, rights, rule, scope and service episode; and
- status, taxonomy, triple, URI, URL and user need.
Their supporting specifications and guidance are annotated in the authoritative-source register.
A–C#
Accessibility#
Design and implementation that allow people with different access needs and assistive technologies to perceive and operate the product.
Active context#
The current overview, search reduction or selected neighbourhood shared across Explorer views.
Adjacency index#
An index that locates relationships next to a selected graph node without loading the whole edge collection.
Agent#
A model combined with instructions, tools, state and an execution loop.
API#
An Application Programming Interface: a machine-oriented contract for one software system to use another.
Artefact#
A produced file or data object, such as a manifest, search shard or validation report.
Approval#
An authorised decision to proceed after reviewing applicable evidence, limitations and residual risk. Approval does not make an unrun check pass.
Assertion#
A statement represented in the data, ideally with status and evidence.
Attestation#
Evidence from a declared process that a computation or material satisfied a contract. A declaration is not itself a passing verdict.
Attribution#
Visible acknowledgement of a source and applicable rights statement. For a map, attribution applies to every visible geometry and basemap layer rather than only to the pack as a whole, including downloadable or offline artefacts.
Bundle or pack#
A published OKF collection and its supporting metadata and indexes.
BFC, BFE, BGC, BSC and BUC#
Office for National Statistics boundary-detail abbreviations. BFC is a
full-resolution clipped boundary; BFE is a full-resolution
extent-of-the-realm boundary; BGC is a generalised 20-metre clipped
boundary; BSC is a super-generalised 200-metre clipped boundary in products
that supply it; and BUC is an ultra-generalised 500-metre clipped boundary.
BFE changes coastal extent, while the other codes combine clipping with
different generalisation levels. The code does not, by itself, identify the
geography or its vintage. See the
boundary-pack proposal.
Boundary generalisation#
Controlled simplification of a boundary's geometry for a declared display scale. It can reduce transfer and rendering cost, but the simplified shape is not a substitute for exact geometry in legal or analytical work. See use the right detail for the scale.
Boundary variant#
A declared treatment of a geography's edge, such as a coastline-clipped or extent-of-the-realm boundary. It is separate from the boundary's vintage and level of detail.
Boundary vintage#
The validity or as-at date that a boundary represents. It remains separate from source publication or version, upstream modification time, retrieval time and pack-build time. See the boundary-pack manifest and join.
Cache#
Stored data reused to avoid repeating a fetch or calculation. Browser caches can be evicted, so cached does not automatically mean available offline.
Candidate#
One proposed set of source, generated and release bytes being tested and reviewed. It is not yet a final release.
Cardinality#
In facets, the number of distinct values. In validation, the permitted number of values for a property.
Chunk#
A bounded group of records stored as one artefact.
Class#
A category of things in a graph vocabulary.
CLML#
Crown Legislation Markup Language, the structured XML used by legislation.gov.uk.
Closed-world validation#
Checking a particular publication as complete under declared rules.
Compression#
Encoding bytes more compactly for storage or transfer.
Concept scheme#
A governed collection of concepts and their labels and relationships, often represented with SKOS.
Conformance#
Meeting the requirements of a named standard or profile.
Context, JSON-LD#
A mapping from short JSON terms to semantic identifiers.
Content-addressed#
Stored or referred to by a digest calculated from the exact bytes. If the bytes change, their content address changes too. See safe caching and failure.
Control plane#
Descriptor and manifest information that tells the application what data artefacts exist and how to access them.
Corpus#
The collection of records under study.
CORS#
Browser rules and response headers that control whether one web origin may read resources from another. A cross-origin range-loaded pack also needs readable range responses and stable object identity.
CRS#
Coordinate Reference System: the frame that gives coordinate numbers spatial meaning. A geometry pack records both its source CRS and each output CRS or tile matrix, together with the exact transformation.
D–H#
Data plane#
The substantial record, relationship, resource and search artefacts described by the control plane.
DCAT#
An RDF vocabulary for catalogues, datasets, data services and distributions.
DCAT-AP#
A European application profile that states community requirements for DCAT metadata.
Descriptor#
The small entry-point document for a large OKF corpus. A proposed geometry-pack reference binds the manifest schema, media type, exact byte count and detached digest before the manifest is parsed.
Deterministic#
Producing the same ordered result from the same inputs and declared rules.
Digest#
The value produced by applying a named hash algorithm to exact bytes. A consumer recalculates it to check that fetched content matches the expected content. A parent descriptor, not the manifest itself, binds a manifest's detached digest. See hash and the PMTiles range-integrity boundary.
Domain profile#
A bounded, validated handoff recording domain research decisions for a later OKF build.
Distribution#
In DCAT, an accessible representation of a dataset, such as a file or API.
DQV#
Data Quality Vocabulary, used to describe quality metrics and annotations.
Edge#
A graph relationship with a source and target.
Entailment#
A statement that follows logically from declared semantic statements and rules.
Entity#
An identified thing. In PROV, an item that participates in provenance.
Facet#
A dimension and its value counts used to explore and filter a result set.
Federation#
An overview-first discovery publication describing independent child bundles and the explicit routes by which they can be loaded.
Fixture#
Controlled input stored for repeatable tests.
Focus node#
The selected graph node around which a bounded context is displayed.
Frontmatter#
Structured metadata at the beginning of a Markdown file.
GeoJSON#
A JSON format for geographic features and geometries.
Geometry representation#
One particular drawing or encoding of a semantic geography, with its own identifier, format, Coordinate Reference System, level of detail, source item and dates, boundary variant, derivation, digest and rights. Replacing a representation does not create a new geography.
Gate#
A named release question with declared evidence and a pass criterion. A gate number has meaning only within its named gate catalogue.
Gate catalogue#
The versioned list that defines gate identifiers, titles, required evidence, pass criteria and responsible roles.
Graph#
Nodes connected by edges.
GSS code#
A Government Statistical Service code used to identify a UK statistical
geography. A safe geometry join also records the geography type and boundary
vintage instead of relying on the code or name alone.
Where published, the canonical statistical-geography IRI retains its specified
http form; a builder must not invent a GSS code.
Hash#
A fixed-length fingerprint calculated from bytes, used for integrity checks.
Hard failure#
A condition the release contract says must block publication. It cannot be turned into a pass by owner acceptance.
Hydration#
Loading fuller data for a compact result or selected record.
I–P#
IDF#
Inverse document frequency, a search weighting that values uncommon terms more than terms appearing in most documents.
Index#
Precomputed data that makes a question such as term lookup or adjacency faster.
Independent reviewer#
A reviewer sufficiently separate from the implementation or expectations under review to reduce self-confirmation. The receipt records the actual role and limitations of that independence.
Inference#
Deriving additional statements from declared rules.
Instance#
A particular member of a class.
Integrity#
Assurance that fetched bytes match an expected hash or signed publication. It does not, by itself, prove publisher authenticity or factual truth.
IRI#
Internationalized Resource Identifier, used to identify semantic resources and properties.
JSON#
A strict text format for objects, arrays and primitive values.
JSON-LD#
JSON for Linked Data, using contexts and identifiers to express an RDF graph.
JSON Schema#
A language for validating the structure of JSON values.
Join table#
An explicit mapping between two identity systems. A boundary-pack join maps one semantic geography to one logical source feature before tiling; repeated fragments of that feature in several tiles are not duplicate joins.
Knowledge graph#
Identified entities and meaningfully typed relationships represented as a graph.
Lazy loading#
Deferring an artefact fetch until it is needed.
Level of detail, LOD#
One of several representations of the same geometry, prepared with different amounts of detail for particular map scales or zoom ranges. Selecting a lower detail for overview display does not change the geography's identity. See the proposed display policy. Each LOD records its output CRS or tile matrix, zoom range, error or tolerance and units, quantisation and topology method, counts, bounds and topology checks.
Literal#
A value such as a string, number or date in an RDF statement.
Manifest#
An inventory of publication artefacts and their locations or integrity data.
Markdown#
A plain-text authoring format with headings, links and lightweight notation.
Materialisation#
Calculating inferred statements in advance and publishing them as derived data.
Material, release#
A file or byte artefact whose identity and digest form part of build or acceptance evidence.
Metadata#
Structured information describing a record, artefact or publication.
Monolithic bundle#
A bundle containing all records and relationships in one file.
Node#
A thing represented in a graph.
Normalisation#
Mapping source variation into a documented common representation while preserving provenance.
Ontology#
A formal model of classes, properties and semantic relationships that may support inference.
OpenAPI#
A machine-readable description of HTTP API operations, inputs, responses and security schemes.
Open-world assumption#
The rule that a statement missing from a graph is generally unknown, not false.
Owner, project or release#
The person or organisation authorised to decide scope, public claims, exceptions, residual-risk acceptance and release promotion.
OWL#
Web Ontology Language, used for expressive class and property semantics.
Posting#
An entry connecting a search term or filter value to a document position.
Predicate#
The identified property or relationship in an RDF triple.
Presentation#
How data is arranged and encoded for a user; it does not change semantic meaning.
Profile#
A named set of rules and choices that applies general standards to a publication community.
Progressive loading#
Loading overview, result and full-detail artefacts in useful stages.
Projection#
A task-oriented view derived from a larger graph or corpus.
Provenance#
Evidence about the entities, activities and agents that produced something.
PWA#
Progressive Web App, a web application with installable or offline-oriented browser capabilities.
Q–Z#
RDF#
Resource Description Framework, a graph model built from subject–predicate– object triples.
RDFS#
RDF Schema, which defines basic vocabulary structures such as classes, properties, domain, range and hierarchies.
Record#
One described item and its fields in an OKF corpus.
Receipt#
A durable record connecting a check or review, its evidence and result to exact inputs, tools, candidate bytes and reviewer roles.
Registry#
A discoverable list of bundle entry points.
Release candidate, RC#
The proposed final release bytes deployed for last-mile public validation. Final promotion should reuse these exact bytes.
Release root#
A digest calculated from an ordered inventory of governed release materials, used to identify the exact candidate covered by evidence and approval.
Relationship#
A directed connection between records or concepts.
Residual risk#
Risk remaining after controls and checks. An authorised owner may accept it without claiming that missing evidence or an absent audit exists.
Resource#
Depending on context, an identified semantic thing or an access/download item attached to a catalogue record.
Retrieval#
Selecting evidence relevant to a task through search, filters, relationships and context assembly.
Route#
A stable application path identifying a record or view state.
Schema#
A formal description of allowed data structure.
SBOM#
Software Bill of Materials, a machine-readable inventory of components and dependencies included in a software release.
Semantic web#
Standards for publishing identified, machine-interpretable linked data.
Semantic geography#
An identified place or area about which assertions are made, independent of the particular files, polygons or tiles used to draw it. It can retain a canonical GSS IRI or another declared source identifier.
Serialisation#
A file syntax used to write an abstract data model, such as JSON-LD or Turtle for RDF.
Sidecar#
A separate supporting file or package referenced by a primary bundle. A sidecar can carry optional, specialised material such as display geometry without making that material part of the bundle's semantic core. See the boundary-pack proposal.
Shard#
A deterministic partition of an index or corpus selected from a key.
SHACL#
Shapes Constraint Language, used to validate RDF graphs.
SKOS#
Simple Knowledge Organization System, used for governed concepts, labels and broader/narrower relationships.
Snapshot#
A coherent publication state captured at a stated version or time.
Source of truth#
The authoritative editable input from which generated artefacts are rebuilt.
Source family#
A researched class of sources in a federation, which may or may not yet have an implemented child bundle.
Static application#
A browser application served as prebuilt files without a custom application server handling each interaction.
Svelte#
The component framework used by the canonical Explorer.
Taxonomy#
A classification, commonly organised as a hierarchy.
Token, search#
A normalised unit of text used by a search index.
Token, security#
A credential representing authority. This is a different meaning from a search token or model token.
Triple#
An RDF statement with subject, predicate and object.
Truncation#
A declared limit that prevented retrieval or preview from processing every possible result.
Turtle#
A compact text serialisation for RDF graphs.
TypeScript#
JavaScript with development-time types, used by the canonical Explorer.
Validation#
Checking data against declared rules.
Waiver#
An authorised, recorded exception to a non-hard requirement. A waiver does not apply to a declared hard failure unless the governing contract itself is changed and re-reviewed.
Vocabulary#
A governed set of identified terms and descriptions.
Web Worker#
A browser background execution context used to keep large search work off the interface thread.
YAML#
A human-oriented structured data notation used in frontmatter and semantic source files.
YAML-LD#
YAML-authored linked data that is parsed under constrained rules and projected to JSON-LD in this repository.
Continue Learning#
Return to the learning-path index or use the repository's more specialised AI infrastructure glossary.