• Recap from the ontology intro & DL lecture

• OWL
– What does the OWL language contain?
– Common misconceptions
• How to model OWL ontologies using Protégé
Components
• concepts = classes in OWL
– represent a set or class of entities in a domain

• instances = individuals in OWL

– represent the members of a set

• relations = (binary) properties in OWL

• axioms
 Example (reminder)

Employment
Person Article

rdf:type rdf:type
rdf:type Ann M. Research at LiU 2016-17
Hans
P.
Information off. Universitetsnytt 22/4 2017
Lecturer Eva B.

Press contact Ontology design patterns

wroteArticle

hasEmployment
 Example (reminder)

Employment

rdfs:subClassOf
Person Article
Admin.
employment rdf:type rdf:type
Res.
employment Ann M. Research at LiU 2016-17
rdf:type Hans
P.
Information off. Universitetsnytt 22/4 2017
Lecturer Eva B.

Press contact Ontology design patterns

wroteArticle
rdf:type
hasEmployment
rdf:type Research = article written by a
article researcher
= person with a
Researcher
research employment
• The most common use case for ontologies on the
Semantic Web
– Classes that represent the types of instances in the
data
– Properties that can be used as predicates in our
RDF triples
Then we can add more... IF we need to
• In this course we focus on task-oriented ontologies,
i.e. ontologies that serves a specific purpose, fulfils a
certain set of requirements
• RDF gives a data representation format and ways to
serialize, but it does not give any special meaning to
vocabulary such as “subClassOf” or “range”

• Triple interpretation is an arbitrary binary relation

• RDF Schema extends RDF with a modelling

vocabulary
– Classes as types for individuals: rdfs:Class, rdfs:Literal,
rdfs:Datatype, rdf:type and rdfs:subClassOf, etc.
– Properties as relations: rdf:Property, rdfs:subPropertyOf,
rdfs:range, rdfs:domain, etc.
– Annotations: rdfs:label, rdfs:comment, etc.
• No distinction between classes and instances
(individuals)
• Properties can themselves have properties
• No distinction between language constructors and
ontology vocabulary, so constructors can be applied
to themselves/each other
• Ability to use simple schema/vocabularies when
describing our resources

• Consistent vocabulary use and sharing

• Simple inference, e.g. inheritance in a taxonomy

• But...
– In some cases too weak to describe resources in
sufficient detail
– Not formally based on any logic
• DL-based language
• OWL2 is the latest version of the standard (https://www.w3.org/TR/owl2-primer/ )
• Different language profiles exist
– OWL EL
• Intended for large ontologies with many classes, mainly used for
classification tasks
• Example use: biomedical ontologies
• Not allowed: negation, disjunction, inverse properties, universal
quantification on properties
– OWL QL
• Covers most features of UML and ER-models, so is suitable for use with
relational data
• Example use: ontologies used to access relational data
– OWL RL
• Reasoning can be implemented as rules
• Does not allow expressions that assume an anonymous individual
• Abstract syntax
– Used in the definition of the language
• Manchester syntax
• “OWL in RDF”
– RDF/XML presentation
– Turtle
• ...
 Example, Example,
Constructor
Turtle syntax Manchester syntax
<Classes> :Human rdf:type owl:Class Class: Human

intersectionOf owl:intersectionOf ( :Human :Male ) Human and Male

unionOf owl:unionOf ( :Male :Female ) Female or Male

complementOf owl:complementOf ( :Male ) not Male

oneOf owl:oneOf ( :John :Mary ) {John, Mary}

 Example, Example,
Axiom
Turtle syntax Manchester syntax
Individual: Mary
<Individual> :Mary rdf:type :Human
Types: Human
Individual: John
<Fact> :Mary :worksWith :John
Facts: worksWith Mary
Individual: Mary
differentFrom :Mary owl:differentFrom :John
DifferentFrom: John
Individual: Mary
sameAs :Mary owl:sameAs :May
SameAs: May
 Example, Example,
Axiom
Turtle syntax Manchester syntax
Class: Woman
subClassOf :Woman rdfs:subClassOf :Human
SubClassOf: Human
Class: Person
equivalentClass :Person owl:equivalentClass :Human
EquivalentTo: Human
[] rdf:type owl:AllDisjointClasses ; DisjointClasses:
disjointClass
owl:members ( :Woman :Man ) . Woman, Man
 Example, Example,
Constructor
Turtle syntax Manchester syntax
owl:onProperty :hasChild ;
someValuesFrom hasChild some Male
owl:someValuesFrom :Male
owl:onProperty :hasChild ;
allValuesFrom hasChild only Female
owl:allValuesFrom :Female
owl:minQualifiedCardinality "2"^^xsd:nonNegativeInteger ;
minCardinality owl:onProperty :hasChild
hasChild min 2

owl:maxQualifiedCardinality "2"^^xsd:nonNegativeInteger ;
maxCardinality owl:onProperty :hasChild
hasChild max 2
 Example, Example,
Axiom
Turtle syntax Manchester syntax
ObjectProperty: hasSon
subPropertyOf :hasSon rdfs:subPropertyOf :hasChild
SubPropertyOf: hasChild

ObjectProperty: hasChild
domain :hasChild rdfs:domain :Parent
Domain: Parent

ObjectProperty: hasSon
range :hasSon rdfs:range :Man
Range: Man

ObjectProperty: worksWith
symmetric :worksWith rdf:type owl:SymmetricProperty
Characteristics: Symmetric

:hasAncestor rdf:type ObjectProperty: hasAncestor

transitive
owl:TransitiveProperty Characteristics: Transitive

ObjectProperty: hasParent
inverseOf :hasParent owl:inverseOf :hasChild
InverseOf: hasChild
• XML namespaces and prefixes
– Turtle: @prefix : <http://example.com/owl/families/> .
@prefix owl: <http://www.w3.org/2002/07/owl#> .
– Manchester: Prefix: : <http://example.com/owl/families/>
Prefix: owl: <http://www.w3.org/2002/07/owl#>
• Datatype properties and XML schema datatypes
– Turtle: :John :hasAge 33
– Manchester: Individual: John
Facts: hasAge "33"^^xsd:integer
• Property chains and keys
• owl:imports
• owl:Ontology
• Annotation properties
– rdfs:label, rdfs:comment, ...
• Disjointness
• Properties do not "belong" to classes
• Interpreting domain and range
• And and or
• Quantification
• Closed and open worlds
• By default, primitive classes are not disjoint.
• Unless we explicitly say so, the description (Animal
and Vegetable) is not an unsatisfiable class
• Similarly with individuals – the so-called Unique
Name Assumption does not hold, and individuals are
not considered to be distinct unless explicitly
asserted to be so.
• Unlike frame-based languages, UML and many other
common modelling languages, in OWL properties do
not "belong" to any specific class
• To "connect" a property to a class we can
– Add domain and range axioms of the property
– Add restrictions on the class
• But neither is necessary for it to be a valid OWL
ontology!
• Note domain and range are NOT interpreted as a
constraint as you might expect
• Domain and range assertions allow us to make
inferences about individuals
• Example
:hasChild rdfs:domain :Parent
:Mary :hasChild :Bob
– If we haven’t said anything else about Mary or
Bob, this is not an error. But we can now infer that
Mary is a Parent
• The logical connectives and and or often cause confusion
– Milk and sugar? Tea or coffee? – think carefully of the meaning
when modeling
– intersection vs. union
• Quantification can be contrary to our intuition.
– Universal quantification over an empty set is true
– :John may belong to the class :OnlyDaughterParent if he
has no child at all and we describe that class as:
:OnlyDaughterParent rdf:type owl:Class ;
owl:equivalentClass [
rdf:type owl:Restriction ;
owl:onProperty :hasChild ;
owl:allValuesFrom :Female
] .
– Existential quantification may imply the existence of an individual
that we don’t know the name of (c.f OWA on next slide)
• The standard semantics of OWL makes an Open World
Assumption (OWA)
– We cannot assume that all information is known about
all the individuals in a domain
– Negation only through contradiction
• Anything might be true unless it can be proven
false

• Closed World Assumption (CWA)

– Named individuals are the only individuals in the
domain
– Negation as failure
• If we don't know that x is of type C, then we
assume that x is NOT of type C