RDF and Ontologies (OWL)

---

1. Anatomy of an Ontology

An ontology structures a domain of discourse into five fundamental components:

Component	Technical Role	Concrete Example
Classes (Concepts)	Abstract categories or types of entities	Person, Faculty, UniversityCourse
Individuals (Instances)	Real-world objects belonging to classes	ex:ram, ex:tcs, ex:cs601
Object Properties	Directed binary relationships between two individuals	ex:enrolledIn, ex:worksFor, ex:teaches
Datatype Properties	Attributes linking individuals to literal values	ex:hasAge (xsd:integer), ex:fullName (xsd:string)
Axioms / Rules	Logical constraints and assertions that must hold true	Every Professor is a Faculty; A person cannot be their own parent

---

2. Resource Description Framework (RDF)

RDF is the fundamental W3C graph data model for representing information on the Web. Every atomic statement is expressed as a triple:

Triple Node Classification

• Subject: The entity being described. Must be an IRI/URI or a Blank Node (anonymous local resource).
• Predicate: The property or relationship. Must always be an IRI/URI.
• Object: The value or target entity. Can be an IRI/URI, a Blank Node, or an RDF Literal (typed or language-tagged string).

---

3. RDF Schema (RDFS) — Vocabulary Modeling

While raw RDF allows any subject to be linked to any object, RDFS (RDF Schema) introduces semantic schema constraints and taxonomic classification:

Core RDFS Constructs & Inference Rules

RDFS Construct	Semantic Meaning	Formal Inference Rule
rdfs:subClassOf	Transitive class specialization	$\text{If } x \in C \land C \sqsubseteq D \implies x \in D$
rdfs:subPropertyOf	Property specialization	$\text{If } \langle x, p, y \rangle \land p \sqsubseteq q \implies \langle x, q, y \rangle$
rdfs:domain	Constrains the type of the Subject	$\text{If } \langle x, p, y \rangle \land \text{domain}(p) = C \implies x \in C$
rdfs:range	Constrains the type of the Object	$\text{If } \langle x, p, y \rangle \land \text{range}(p) = D \implies y \in D$

---

4. Web Ontology Language (OWL 2)

RDFS cannot express negation, disjointness, cardinality, or property characteristics. OWL (Web Ontology Language) extends RDFS with expressive Description Logics ($\mathcal{SROIQ}^{(D)}$).

Key Expressive Capabilities of OWL

1. Class Constructors: Union ($\sqcup$), Intersection ($\sqcap$), Complement ($
   eg$), Enumeration (oneOf).
2. Property Characteristics:
   • Transitive ($P(a, b) \land P(b, c) \implies P(a, c)$) e.g., ex:isAncestorOf.
   • Symmetric ($P(a, b) \implies P(b, a)$) e.g., ex:isSiblingOf.
   • Functional ($P(a, b) \land P(a, c) \implies b = c$) e.g., ex:hasBiologicalMother.
   • Inverse ($P(a, b) \iff Q(b, a)$) e.g., ex:employs vs ex:worksFor.
3. Property Restrictions:
   • Existential ($\exists$): someValuesFrom (at least one value must belong to class $C$).
   • Universal ($\forall$): allValuesFrom (all values, if any, must belong to class $C$).
   • Cardinality: minCardinality, maxCardinality, exactCardinality.

---

5. OWL 2 Profiles & Species (Expressiveness vs Decidability)

W3C standardizes three lightweight profiles designed for polynomial-time performance, alongside full Description Logic:

Profile / Species	Underlying Logic	Computational Complexity	Ideal Real-World Use Case
OWL 2 EL	$\mathcal{EL}^{++}$ (Existential)	Polynomial time $O(n^k)$	Large biomedical ontologies (e.g., SNOMED CT, Gene Ontology with 400,000+ classes).
OWL 2 QL	DL-Lite (Query Language)	LogSpace (in data)	Relational database mapping & query rewriting without materializing triples.
OWL 2 RL	Rule-extended Description Logic	Polynomial time	Rule-based reasoning engines executed directly inside scalable RDF triple stores.
OWL 2 DL	$\mathcal{SROIQ}^{(D)}$	NEXPTIME-complete	Maximum expressiveness while strictly preserving decidability and termination.
OWL 2 Full	Unrestricted RDF graph logic	Undecidable	Complete freedom of expression, but algorithms cannot guarantee termination.

---

6. Multi-Format Ontology Modeling Snippets

The following tabs contrast how a university ontology with restrictions is declared across different syntax serializations:

@prefix owl:  <http://www.w3.org/2002/07/owl#> .
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
@prefix ex:   <http://example.org/university#> .

# Class: Student is a subclass of Person
ex:Student a owl:Class ;
    rdfs:subClassOf ex:Person ;
    rdfs:subClassOf [
        a owl:Restriction ;
        owl:onProperty ex:enrolledIn ;
        owl:someValuesFrom ex:Course
    ] .

# Disjoint Classes: Faculty and Student cannot overlap
[ a owl:AllDisjointClasses ;
  owl:members ( ex:Student ex:Faculty )
] .

# Property Characteristics: transitive prerequisite
ex:hasPrerequisite a owl:TransitiveProperty .

---

7. Step-by-Step Ontology Engineering Methodology

Building industrial-strength ontologies requires a structured lifecycle (such as the Grüninger-Fox or Methontology framework):

---

8. Deep Dive: Semantic Reasoners

---