SKOS (Simple Knowledge Organization System)

SKOS (Simple Knowledge Organization System) is a W3C standard for representing structured vocabularies, taxonomies, thesauri, and classification schemes using RDF. SKOS provides a lightweight ontology for organizing concepts with labels, hierarchies, relationships, and documentation, making knowledge organization systems machine-readable and interoperable.

Core SKOS Elements

Concepts and Concept Schemes

Concepts are the fundamental units (topics, categories, subjects):

@prefix skos: <http://www.w3.org/2004/02/skos/core#> .
@prefix ex: <http://example.org/topics/> .

# Concept Scheme (container for concepts)
ex:SubjectScheme a skos:ConceptScheme ;
    skos:prefLabel "Subject Classification" ;
    skos:definition "Topics for organizing documents" .

# Individual concepts
ex:Technology a skos:Concept ;
    skos:inScheme ex:SubjectScheme ;
    skos:prefLabel "Technology"@en ;
    skos:prefLabel "Technologie"@fr ;
    skos:definition "Applied sciences and engineering" .

ex:ArtificialIntelligence a skos:Concept ;
    skos:inScheme ex:SubjectScheme ;
    skos:prefLabel "Artificial Intelligence"@en ;
    skos:altLabel "AI"@en ;
    skos:altLabel "Machine Intelligence"@en ;
    skos:definition "Computer systems that perform tasks requiring human intelligence" .

Labels

SKOS provides three types of labels:

Preferred Label (skos:prefLabel)

The primary, canonical name:

ex:MachineLearning skos:prefLabel "Machine Learning"@en .

Rules:

• Exactly one per language
• Used for display and user interfaces
• Should be unique within concept scheme

Alternative Label (skos:altLabel)

Synonyms and variants:

ex:MachineLearning
    skos:prefLabel "Machine Learning"@en ;
    skos:altLabel "ML"@en ;
    skos:altLabel "Statistical Learning"@en ;
    skos:altLabel "Automated Learning"@en .

Use cases:

• Abbreviations
• Synonyms
• Historical terms
• Spelling variations

Hidden Label (skos:hiddenLabel)

For search matching but not display:

ex:MachineLearning
    skos:prefLabel "Machine Learning"@en ;
    skos:hiddenLabel "machinelearning"@en ;  # No space
    skos:hiddenLabel "machin learning"@en ;  # Common misspelling
    skos:hiddenLabel "machine lerning"@en .  # Another typo

Use cases:

• Common misspellings
• Outdated terms
• Indexing terms

Hierarchical Relations

Broader/Narrower

Express taxonomic hierarchies:

ex:Technology a skos:Concept ;
    skos:prefLabel "Technology" .

ex:ArtificialIntelligence a skos:Concept ;
    skos:prefLabel "Artificial Intelligence" ;
    skos:broader ex:Technology .  # AI is narrower than Technology

ex:MachineLearning a skos:Concept ;
    skos:prefLabel "Machine Learning" ;
    skos:broader ex:ArtificialIntelligence .

ex:DeepLearning a skos:Concept ;
    skos:prefLabel "Deep Learning" ;
    skos:broader ex:MachineLearning .

Inverse relationship:

ex:Technology skos:narrower ex:ArtificialIntelligence .
ex:ArtificialIntelligence skos:narrower ex:MachineLearning .

Hierarchy:

Technology
  └─ Artificial Intelligence
       └─ Machine Learning
            └─ Deep Learning

Top Concepts

Mark the root concepts of a hierarchy:

ex:SubjectScheme a skos:ConceptScheme ;
    skos:hasTopConcept ex:Technology ;
    skos:hasTopConcept ex:Science ;
    skos:hasTopConcept ex:Healthcare .

ex:Technology skos:topConceptOf ex:SubjectScheme .

Associative Relations

Related Concepts (skos:related)

Non-hierarchical associations:

ex:MachineLearning a skos:Concept ;
    skos:prefLabel "Machine Learning" ;
    skos:related ex:DataScience ;  # Related but not hierarchical
    skos:related ex:Statistics ;
    skos:related ex:NeuralNetworks .

ex:DataScience skos:related ex:MachineLearning .  # Symmetric

Use cases:

• Cross-domain connections
• See-also references
• Complementary topics

Documentation Properties

Definition

Formal explanation:

ex:MachineLearning
    skos:definition """A subset of artificial intelligence focused on
    algorithms that learn patterns from data without explicit programming."""@en .

Scope Note

Usage guidance:

ex:MachineLearning
    skos:scopeNote """Use for documents about supervised, unsupervised,
    and reinforcement learning. For neural networks specifically, use
    'Deep Learning'."""@en .

Example

Illustrative usage:

ex:MachineLearning
    skos:example "Image recognition systems using convolutional neural networks"@en ;
    skos:example "Spam filters trained on email corpora"@en .

History Note

Provenance and changes:

ex:ArtificialIntelligence
    skos:historyNote """Originally coined in 1956 at the Dartmouth Conference.
    Term usage expanded significantly after 2010."""@en .

Editorial Note

Internal management notes:

ex:MachineLearning
    skos:editorialNote "Consider splitting into supervised/unsupervised subcategories in next version"@en .

Complete Example: Technology Taxonomy

@prefix skos: <http://www.w3.org/2004/02/skos/core#> .
@prefix ex: <http://example.org/tech/> .

# Concept Scheme
ex:TechTaxonomy a skos:ConceptScheme ;
    skos:prefLabel "Technology Taxonomy"@en ;
    skos:definition "Classification system for technology topics" ;
    skos:hasTopConcept ex:Technology .

# Top Concept
ex:Technology a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:topConceptOf ex:TechTaxonomy ;
    skos:prefLabel "Technology"@en ;
    skos:narrower ex:SoftwareEngineering, ex:ArtificialIntelligence .

# Software Engineering Branch
ex:SoftwareEngineering a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:prefLabel "Software Engineering"@en ;
    skos:altLabel "Software Development"@en ;
    skos:broader ex:Technology ;
    skos:narrower ex:WebDevelopment, ex:MobileDevelopment ;
    skos:related ex:ArtificialIntelligence .

ex:WebDevelopment a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:prefLabel "Web Development"@en ;
    skos:altLabel "Web Dev"@en ;
    skos:broader ex:SoftwareEngineering ;
    skos:definition "Building applications for the World Wide Web" .

# AI Branch
ex:ArtificialIntelligence a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:prefLabel "Artificial Intelligence"@en ;
    skos:altLabel "AI"@en ;
    skos:broader ex:Technology ;
    skos:narrower ex:MachineLearning, ex:NaturalLanguageProcessing ;
    skos:related ex:SoftwareEngineering ;
    skos:definition "Computer systems that simulate human intelligence" .

ex:MachineLearning a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:prefLabel "Machine Learning"@en ;
    skos:altLabel "ML"@en ;
    skos:altLabel "Statistical Learning"@en ;
    skos:broader ex:ArtificialIntelligence ;
    skos:narrower ex:DeepLearning, ex:ReinforcementLearning ;
    skos:definition "Algorithms that learn from data" ;
    skos:example "Image classification, recommendation systems" .

ex:DeepLearning a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:prefLabel "Deep Learning"@en ;
    skos:altLabel "Deep Neural Networks"@en ;
    skos:broader ex:MachineLearning ;
    skos:definition "Neural networks with multiple layers" .

ex:NaturalLanguageProcessing a skos:Concept ;
    skos:inScheme ex:TechTaxonomy ;
    skos:prefLabel "Natural Language Processing"@en ;
    skos:altLabel "NLP"@en ;
    skos:broader ex:ArtificialIntelligence ;
    skos:related ex:MachineLearning .

SKOS Mapping Properties

Link concepts across different vocabularies:

Exact Match

Concepts are equivalent:

ex:ArtificialIntelligence skos:exactMatch dbpedia:Artificial_intelligence .

Close Match

Concepts are closely related but not identical:

ex:MachineLearning skos:closeMatch wikidata:Q2539 .

Broader/Narrower Match

Hierarchical relationships across schemes:

ex:DeepLearning skos:broadMatch dbpedia:Machine_learning .
ex:Technology skos:narrowMatch dbpedia:Software_engineering .

Related Match

Associative relationships across schemes:

ex:AI skos:relatedMatch dbpedia:Cognitive_science .

SKOS in TrustGraph

Loading SKOS Taxonomies

# Load SKOS taxonomy
tg-load-text \
  --text "$(cat tech-taxonomy.ttl)" \
  --title "Technology Taxonomy"

# Query using taxonomy
tg-invoke-graph-rag \
  -q "Find all concepts related to Artificial Intelligence"

Querying SKOS with SPARQL

PREFIX skos: <http://www.w3.org/2004/02/skos/core#>

# Find all narrower concepts (subcategories)
SELECT ?concept ?label
WHERE {
  ?concept skos:broader ex:ArtificialIntelligence ;
           skos:prefLabel ?label .
}

# Find all synonyms for a term
SELECT ?altLabel
WHERE {
  ex:MachineLearning skos:altLabel ?altLabel .
  FILTER(lang(?altLabel) = "en")
}

# Navigate full hierarchy
SELECT ?narrower ?label
WHERE {
  ex:Technology skos:narrower+ ?narrower .
  ?narrower skos:prefLabel ?label .
}

Use Cases

1. Document Classification: Categorize content using controlled vocabularies
2. Search Enhancement: Query expansion with synonyms and related terms
3. Data Integration: Map concepts across different systems
4. Semantic Navigation: Browse content hierarchically
5. Multilingual Support: Maintain labels in multiple languages
6. Thesaurus Management: Manage controlled vocabularies
7. Faceted Search: Organize search filters by taxonomy

SKOS vs OWL

When to use SKOS:

• Organizing content for users
• Managing controlled vocabularies
• Simple hierarchical classification
• Multilingual label management

When to use OWL:

• Formal reasoning required
• Data validation against constraints
• Complex domain modeling
• Automated inference

Best Practices

1. One Language, One Label: Use @en, @fr language tags consistently
2. Unique Preferred Labels: Avoid duplicate prefLabels in same scheme
3. Bidirectional Relations: Define both broader and narrower for clarity
4. Document Decisions: Use scopeNote and editorialNote
5. Hierarchies: Keep reasonably shallow (3-5 levels max)
6. Related vs Broader: Use skos:related for non-hierarchical associations
7. Multilingual: Provide labels in all relevant languages
8. Versioning: Track changes to concept schemes over time

Limitations

1. No Formal Semantics: Cannot express logical constraints
2. Weak Validation: No strict enforcement of rules
3. Ambiguous Relations: broader/narrower less precise than OWL subClassOf
4. No Inference: Limited reasoning compared to OWL
5. Flat Properties: Cannot define property hierarchies

See Also

• Taxonomy - Hierarchical classifications
• Vocabulary - Standardized term sets
• RDF - Foundation for SKOS
• OWL - More expressive ontology language
• Semantic Web - Machine-readable web vision
• Ontology - Formal domain specifications