RDFS (RDF Schema)

These are the foundational constructs for defining categories of things. rdfs:Class is used to declare a class (a category or type). rdfs:Resource is the class of everything; all things described in RDF are instances of rdfs:Resource. Every user-defined class is a subclass of rdfs:Resource.

• Example: ex:Person a rdfs:Class . declares ex:Person as a class.
• Example: ex:Alice a ex:Person . states that the individual ex:Alice is an instance (a member) of the class ex:Person.

This property establishes a hierarchical taxonomy between classes. It states that the members (instances) of one class are also members of another, more general class. This enables inheritance; any property or characteristic defined for the superclass can be inferred for its subclasses.

• Example: ex:Employee rdfs:subClassOf ex:Person .
• Inference: If ex:Bob a ex:Employee . is stated, a reasoner can infer ex:Bob a ex:Person .
• The relationship is transitive: if A is a subclass of B, and B is a subclass of C, then A is a subclass of C.

Analogous to rdfs:subClassOf, this property creates a hierarchy for properties (relationships). It indicates that if a pair of resources is connected by the sub-property, they are also connected by the super-property. This is key for organizing and specializing relationships.

• Example: ex:manages rdfs:subPropertyOf ex:knows .
• Inference: If ex:Carol ex:manages ex:Team . is stated, it can be inferred that ex:Carol ex:knows ex:Team .
• This supports property specialization, allowing for more precise relationship definitions while maintaining connection to broader concepts.

These properties constrain the types of resources that can serve as the subject or object of a given property, defining a simple form of schema or type checking.

• rdfs:domain specifies the class of the subject (the resource that has the property). If a property P has domain D, and a triple X P Y exists, then X is inferred to be of type D.
  • Example: ex:hasEmployee rdfs:domain ex:Company .
• rdfs:range specifies the class of the object (the value of the property). If a property P has range R, and a triple X P Y exists, then Y is inferred to be of type R.
  • Example: ex:hasEmployee rdfs:range ex:Person .

These properties provide human-readable documentation for resources, which is critical for ontology usability and maintenance. They do not affect machine reasoning.

• rdfs:label provides a human-readable name for a resource. It is often used with language tags for internationalization.
  • Example: ex:Person rdfs:label "Person"@en .
  • Example: ex:Person rdfs:label "Personne"@fr .
• rdfs:comment provides a longer description or explanation of a resource's meaning and intended use.
  • Example: ex:subClassOf rdfs:comment "The subject is a subclass of a class."@en .

RDFS provides a basic framework for representing collections or groups of resources, though it is less expressive than more modern approaches like RDF Lists.

• rdfs:Container is a superclass for container classes: rdf:Bag (unordered), rdf:Seq (ordered), and rdf:Alt (alternatives).
• rdfs:member is a generic property to relate a container to its members.
• rdf:_1, rdf:_2, ... are properties used to denote the specific position of a member in an rdf:Seq.

Limitation: RDFS does not provide a way to state that a container is closed or to define its cardinality. For more rigorous collection semantics, OWL is required.

RDFS is used to create simple taxonomies by defining classes and organizing them into hierarchies using the rdfs:subClassOf property. This allows for inheritance, where members of a subclass are automatically inferred to be members of the superclass. For example:

• Defining :Employee as a subclass of :Person.
• Any resource typed as an :Employee is automatically inferred to also be a :Person. This hierarchical organization is the backbone of basic ontology engineering and enables categorical reasoning over data.

RDFS allows for the definition and organization of properties (predicates). Using rdfs:subPropertyOf, you can create hierarchies of relationships, enabling more generalized queries. Key applications include:

• Defining :reportsTo as a subproperty of :knows.
• Using rdfs:domain and rdfs:range to specify the class of the subject and object expected for a property. For instance, declaring that the domain of :author is :Document and its range is :Person provides implicit typing and schema validation for triples.

A core use of RDFS is to enable forward-chaining inference within a triplestore. By applying the formal semantics of RDFS, systems can automatically deduce new triples. Common inferences include:

• Type propagation via rdfs:subClassOf.
• Property inheritance via rdfs:subPropertyOf.
• Domain/Range typing: If a triple (X, P, Y) exists and P has a domain D and range R, the system can infer (X, rdf:type, D) and (Y, rdf:type, R). This provides a basic level of semantic reasoning without the complexity of OWL.

RDFS serves as the entry point for ontology development, providing a simpler alternative to OWL for projects that require structured vocabularies but not complex constraints. It is ideal for:

• Defining a controlled vocabulary for data annotation (semantic tagging).
• Creating a shared data model for integrating heterogeneous sources in a data lake or warehouse.
• Providing the schema layer for a knowledge graph, where rdfs:label and rdfs:comment are used extensively for human-readable labels and definitions.

RDFS vocabularies act as a canonical model to align disparate data sources. By mapping source fields to RDFS classes and properties, you achieve semantic integration. This is foundational for:

• Linked Data initiatives, where publishers use common RDFS vocabularies (like FOAF or Dublin Core) to ensure interoperability.
• Building a semantic data fabric, where RDFS provides the unifying layer of meaning across databases, APIs, and files.
• Enabling federated queries with SPARQL, as queries are written against the unified RDFS schema rather than source-specific structures.

RDFS includes built-in properties specifically designed for documentation, making schemas self-describing. This is critical for data governance and collaboration.

• rdfs:label provides a human-readable name for a resource.
• rdfs:comment offers a full description or definition.
• rdfs:seeAlso and rdfs:isDefinedBy link to related documentation. These properties transform a machine-oriented schema into a living documentation system that supports both developers and domain experts, improving the usability and adoption of the semantic model.

The foundational data model for the Semantic Web upon which RDFS is built. RDF represents information as a directed graph of subject-predicate-object triples. It provides the universal syntax for expressing data, while RDFS adds semantic meaning by defining vocabularies for classes and properties.

• Core Construct: A triple (<subject> <predicate> <object>).
• Serialization Formats: Includes Turtle, RDF/XML, and JSON-LD.
• Foundation: All RDFS statements are themselves valid RDF triples.

A more expressive family of knowledge representation languages that builds upon RDFS. While RDFS defines basic taxonomies, OWL adds formal semantics based on Description Logics for defining complex ontologies with precise constraints.

• Key Differentiators: Supports property characteristics (e.g., symmetry, transitivity), class disjointness, and cardinality restrictions.
• Profiles: OWL 2 EL, QL, and RL are standardized subsets optimized for specific reasoning tasks.
• Use Case: Used when rigorous, automated logical inference is required beyond simple hierarchy management.

A W3C standard for validating RDF data against a set of conditions. While RDFS defines a schema for inference, SHACL defines a constraint language for data quality assurance.

• Core Function: Validates that an RDF graph conforms to a set of "shapes" specifying required properties, data types, and value ranges.
• Complementary Role: Often used alongside RDFS/OWL; the ontology defines meaning, SHACL validates instance data integrity.
• Example Constraint: Ensuring all ex:Employee instances have exactly one ex:employeeID that is a string.

An RDF-based vocabulary for representing knowledge organization systems like thesauri, classification schemes, and taxonomies. It is simpler and more focused than full RDFS/OWL ontologies.

• Core Concepts: skos:Concept, skos:prefLabel, skos:altLabel, skos:broader, skos:narrower.
• Relationship to RDFS: SKOS concepts can be related to RDFS/OWL classes, but SKOS is optimized for labeling and linking concepts for human navigation.
• Primary Use: Publishing and linking controlled vocabularies on the web.

The standard query language for RDF data. It is used to retrieve and manipulate information stored in any RDF-based system, including those using RDFS vocabularies.

• Query Forms: SELECT (returns tables), CONSTRUCT (creates new RDF graphs), ASK (returns boolean), DESCRIBE (returns relevant triples).
• Leverages RDFS: SPARQL queries can use RDFS-defined classes and properties in their graph patterns. The SPARQL 1.1 standard includes entailment regimes for RDFS reasoning during query execution.
• Endpoint: Accessed via a standardized HTTP protocol called a SPARQL endpoint.

A purpose-built database designed for the efficient storage, retrieval, and management of RDF triples. It is the native storage engine for RDFS-based knowledge graphs.

• Key Features: Optimized for graph pattern matching (SPARQL), often includes built-in RDFS inference engines, and supports named graphs for data partitioning.
• Contrast with Relational DBs: Stores data as subject-predicate-object statements rather than in fixed tables and columns.
• Examples: Open-source systems include Apache Jena Fuseki and RDF4J; commercial offerings include GraphDB and Stardog.