SKOS (Simple Knowledge Organization System)

A Concept is the central class in SKOS, representing a unit of thought or an idea within a knowledge organization system. It is an abstract notion, not a word or term.

• Defined as an instance of skos:Concept.
• Identified by a URI, making it globally referenceable.
• Serves as the anchor point for labels, notes, and semantic relations.
• Example: The abstract idea of "Artificial Intelligence" (identified by a URI like http://example.com/concept/ai) is a SKOS Concept, distinct from the label "AI" used to denote it.

Lexical Labels are the human-readable names attached to a SKOS Concept. SKOS defines specific properties for different types of labels to support multilingual and contextual representation.

• skos:prefLabel: The preferred, primary label for a concept in a given language. A concept can have only one prefLabel per language tag.
• skos:altLabel: An alternative or synonymous label (e.g., an acronym or common misspelling).
• skos:hiddenLabel: A label that may be useful for search and retrieval but is not intended for display (e.g., a common typo).
• All labels are language-tagged strings (e.g., "Machine Learning"@en, "Aprendizaje Automático"@es).

Semantic Relations define the conceptual structure of a vocabulary by linking SKOS Concepts. These relations are distinct from generic RDF links, as they carry specific meaning about conceptual organization.

• skos:broader / skos:narrower: Creates a hierarchical parent-child relationship (e.g., Neural Network is skos:narrower than Machine Learning). These are transitive relations.
• skos:related: Indicates an associative, non-hierarchical link between concepts that are semantically related (e.g., Machine Learning is skos:related to Data Mining).
• skos:topConceptOf: Links a concept to a skos:ConceptScheme, denoting it as a top-level entry point.

A Concept Scheme (skos:ConceptScheme) is a collection or aggregation of one or more SKOS Concepts. It represents an entire controlled vocabulary, thesaurus, or classification scheme as a single, identifiable entity.

• Serves as the root container for a set of related concepts.
• Concepts are linked to their scheme using skos:inScheme.
• Enables the management and publication of distinct, self-contained vocabularies (e.g., a corporate product taxonomy vs. an industry standard classification).
• A concept can be a member of multiple schemes, enabling cross-walking and mapping between different KOS.

Documentation Properties allow for the attachment of human-readable notes and definitions to SKOS Concepts, providing essential context and scope notes for users and maintainers.

• skos:definition: A complete explanation of the meaning of a concept.
• skos:scopeNote: Information about the intended meaning or usage of a concept, clarifying its boundaries.
• skos:example: An example of the concept's use.
• skos:historyNote: Information about the past state/use of the concept.
• skos:editorialNote: Administrative information for vocabulary editors.
• Like labels, notes can be language-tagged to support multilingual documentation.

SKOS is the de facto standard for publishing and interlinking enterprise thesauri and taxonomies as linked data. It models core thesaurus relationships:

• skos:broader / skos:narrower for hierarchical (parent/child) relationships.
• skos:related for associative relationships between concepts.
• skos:prefLabel, skos:altLabel, and skos:hiddenLabel for managing preferred terms, synonyms, and non-preferred terms. This enables consistent tagging, content discovery, and semantic search across digital asset management systems, content management platforms, and data catalogs.

SKOS serves as a lightweight interlingua for aligning disparate classification schemes. Organizations use SKOS mapping properties to integrate data from multiple sources:

• skos:exactMatch, skos:closeMatch, skos:broadMatch, skos:narrowMatch, and skos:relatedMatch. These properties declare the degree of correspondence between concepts from different vocabularies (e.g., mapping an internal product code to an industry-standard classification like UNSPSC). This is foundational for ontology alignment and creating a unified semantic layer in a knowledge graph.

While OWL is used for heavyweight ontologies with complex constraints, SKOS provides an 'ontology lite' approach suitable for many business needs. It offers a pragmatic balance between simplicity and expressiveness for knowledge organization systems (KOS). Key differentiators from OWL:

• SKOS skos:ConceptScheme organizes concepts, but does not enforce strict logical disjointness or equivalence.
• It focuses on lexical labels and documentation notes (skos:definition, skos:scopeNote) rather than formal axioms.
• It is designed for human curation and browsing as much as for machine reasoning.

SKOS-powered vocabularies directly improve search relevance and faceted navigation. By tagging content with concepts from a SKOS vocabulary, systems can:

• Expand queries using skos:altLabel (synonyms).
• Suggest broader or narrower topics during search using skos:broader/skos:narrower.
• Provide consistent filters and facets in user interfaces.
• Enable semantic search that understands conceptual meaning, not just keywords. This is a core component of semantic annotation pipelines.

The foundational data model for the Semantic Web upon which SKOS is built. RDF represents information as a directed graph of subject-predicate-object triples. SKOS concepts, labels, and semantic relations are all expressed as RDF triples, making SKOS vocabularies inherently machine-readable and integrable with other RDF data.

• Example: The triple (ex:Cat, skos:prefLabel, "Felis catus") uses RDF to state a SKOS concept's preferred label.

A semantic extension of RDF that provides a basic vocabulary for defining taxonomies. RDFS introduces fundamental constructs like rdfs:Class, rdfs:subClassOf, and rdfs:domain/rdfs:range. SKOS uses and extends RDFS to define its own classes (like skos:Concept) and properties (like skos:broader). While RDFS is used for class hierarchies, SKOS is optimized for organizing concept schemes and thesauri.

A family of expressive knowledge representation languages based on Description Logics, used to define rich, complex ontologies with formal semantics. OWL enables powerful automated reasoning (e.g., consistency checking, classification). SKOS and OWL serve complementary purposes: SKOS is for lightweight vocabulary sharing, while OWL is for defining precise, logically rigorous ontologies. They can be used together, with OWL ontologies providing a formal backbone for SKOS concept schemes.

The standard query language for RDF data. SPARQL is used to query SKOS vocabularies stored in a triplestore. It enables complex searches across concepts, labels, and relationships.

• Example Query: Find all concepts with a broader term "Artificial Intelligence": SELECT ?concept WHERE { ?concept skos:broader <http://example.org/AI> . }
• Key Use: Federated querying across multiple distributed SKOS vocabularies.

A set of best practices for publishing structured data on the web using URIs, RDF, and HTTP. SKOS is a premier vocabulary for implementing Linked Data principles for controlled vocabularies. By publishing a SKOS concept scheme as Linked Data, each concept gets a dereferenceable HTTP URI, allowing other systems to discover and link to it directly, creating a web of interconnected, reusable knowledge organizations systems.

Traditional knowledge organization systems that SKOS is designed to represent and share digitally.

• Taxonomy: A hierarchical classification system (e.g., a product category tree). SKOS models this with skos:broader/skos:narrower properties.
• Thesaurus: A controlled vocabulary that includes not only hierarchy but also associative relationships ("related term") and synonyms. SKOS models these with skos:related and skos:altLabel. SKOS provides a common, interoperable format for migrating these legacy systems to the web.