Ontologies and OWL
Recap
• Sharing knowledge requires embedding semantics with data
• RDF: graph model for knowledge
• No type, no restrictions on terms
• RDFS: knowledge model extends RDF
• Allows creating vocabularies (terminologies)
• lacks means for details: no restriction on range and domain, no cardinality, other
properties
• SKOS: builds on RDFS/RDF to create structured vocabularies
• Taxonomies, thesauri…
• Have limitations in expressiveness and reasoning
• Need for a language to express strong semantics
Semantic spectrum
Semantics and associated functions
What is an ontology
• Metadata
• Data describing the content and meaning of resources and
services
• Requires parties to use the same terms
• Terminologies
• Shared and common vocabularies
• Used with search engines, software agents, librarians, authors…
• Requires parties to use the same meaning
• Ontologies
• Shared and common knowledge of a domain
• Meant for exchange, search, and discovery
The meaning triangle
Thought
(I would like to drink camel milk)
Symbol Referent C.a.m.e.l
ﺟﻣل
subjective
objective
Ontology
• An explicit specification of a conceptualization (Gruber 1993)
• An ontology describes the common words, concepts and relationships
between concepts used to describe and represent knowledge in a specific
domain
• The concept of ontology refers to different forms of knowledge
representation, with weak to strong semantics
• Taxonomy (structured vocabulary) – relational model
• Thesaurus (words and synonyms) – entity relationship
• Conceptual models (complex knowledge, concepts, properties, rules) - RDF/RDFS
• Logical theory (rich complex consistent knowledge, axioms, inference rules) – OWL,
Description Logic, First order logic
• An ontology is said to be well-formed if it is expressed in well-defined
formalism that can be machine interpretable
Ontology modelling
• Explicit description of a domain, in terms of
• Concepts: classes, sets, types, predicates
• Properties of concepts: attributes (named data properties)
• Relationships: relations between concepts (named object properties)
• Constraints (or axioms): constraints on properties and concepts such as type
of values, range of values, cardinality…
• Values: actual values, strings…
• Individuals (instances): actual things, people, concepts…
• When ontology includes also individuals (actual data), it is called a
knowledge base
Relational Schema vs. Ontology
• Relational schema
• Meant to organise data into databases
• Relationships are implicit and require human or code to do the interpretation
(semantics in the code or in the human mind)
• Without semantics, no human or program can use the data in a meaningful
way
• Ontology
• Meant to share information along with semantics
• Relationships defined formally (based on logical constructs)
• Interpretation is possible by both human and machine
Examples
Simple ontologies with informal representation
Terminology
In semantic computing, we use two conventional terms:
• TBox
• the ontology (knowledge model), with concepts and relationships…
• ABox
• the data instances (individuals) as per the ontology
OWL: Web Ontology Language
A knowledge modelling language whose requirements are
• Extends existing Web standards: XML, RDF, RDFS
• Easy to understand and use: based on familiar KR idioms
• Formally specified - describes the meaning of knowledge precisely
• Having high expressive power
• Provides automated reasoning support
OWL: what for?
• OWL is primarily concerned with defining terminology that can be
used in RDF documents, i.e., classes and properties.
• OWL specifies the terminology in terms of classes and properties ad
their characteristics
• Individuals (instances) can also be defined in an owl document
Example of OWL
It is an RDF document
Metadata about the ontology
Hierarchy of classes
Another example
Data property
Object property
Namespaces
metadata
Object property
Class hierarchy Object property
Object property with its own characteristics
Object property hierarchy
VOWL Representation http://vowl.visualdataweb.org/v2/
Defines a visual language for the user-oriented representation of ontologies.
Example of FOAF
Reasoning with OWL
• Consistency checking: No contradictions
• ABox does not contradict TBox (instances compliant with ontology)
• Concept satisfiability: possibility of adding individuals from class
• Can a class have individuals?
• Classification: create full inheritance hierarchy to answer questions
• Does class A subsumes class B?
• Realisation: compute the direct type (class) of each individual
• What class does individual X belongs to?
• Realisation can be done after classification as we need the full inheritance
hierarchy
Protégé
• Protégé is an open source ontology editor using OWL
• It allows creating through a graphical user interface all components of
the ontology
• Can be downloaded freely from https://protege.Stanford.edu
• It provides a number of views to suit the need of the user at different
stages of the ontology engineering process
1. Creating an ontology
4. Save ontology
1. set IRI 5. Select preferred
format
2. Add metadata
3. Define prefixes
2. Create class hierarchy Use meaningful names
Add annotations to each class
Create a hierarchy
Set any class characteristics,
example disjoint
owl:equivalentClass
• Two classes that refer to the
same things
• Written in Turtle format
:Human owl:equivalentClass :Person
Disjoint classes
• When individuals cannot belong
to two classes, these classes are
declared disjoint
:man owl:disjointWith :Woman ;
• We can also make one class
disjoint from a list of classes, this
gives in turtle (a collection):
[ rdf:type owl:AllDisjointClasses ;
owl:members ( :C1 :C2 :C3 ) ] .
3. Create object properties (relationships)
• Use verbs to avoid ambiguity
with classes
• Add any characteristics
C
:marriedTo a rdf:Property.
• Add :hasHusband C
• Make it inverse of :hasWife
Property hierarchy
• Properties can be in a hierarchy
• Example
:hasWife rdfs:subPropertyOf :hasSpouse .
Domain and range
• Restrict the type of subject and
object for a property
• Example
:hasWife rdf:type owl:ObjectProperty ;
rdfs:subPropertyOf :hasSpouse ;
rdf:type owl:AsymmetricProperty ;
rdfs:domain :Man ;
rdfs:range :Woman.
4. Data properties
• These are attributes to describe
individuals from classes
• Example
5. Add individuals
• Add individuals (instances)
• Define their type (class)
:Mary rdf:type owl:NamedIndividual :Woman ;
• Assert their properties
:John :hasWife :Mary;
• No need to assert that
:Mary :hasHusband :John;
• It will be discovered by
reasoner…
Protégé: using a reasoner
• A reasoner is an algorithm that
applies local inference to
discover
• Inconsistencies (any contradiction)
• Subsumptions (class inheritance)
• Realisations (individuals of classes)
• By applying class and property
characteristics (inverseOf,
symmetric…), find new
relationships
Inferred information in yellow
Property hasWife
Class Man
Individual Mary
Individual John
Principles when creating an ontology
• There exists no unique perfect knowledge model for a domain
• The model should serve your application
• An ontology cannot be created at once
• An iterative process is required to refine the ontology multiple times to
consider various use cases
• Classes and properties should be as close as possible to the
real/physical objects of your domain
• Consider nouns for classes and verbs for properties
Process of creating an ontology
1. Determine scope and domain of ontology
2. Consider reusing existing ontologies
3. Enumerate important terms in the ontology
4. Define the classes and class hierarchy
5. Define the properties of classes
• Properties:
6. Relationships
7. Define restrictions on properties
• Value type (String, Number, Boolean, Enumerated, Instance)
• Value cardinality
8. Define the instances
Ontology Development Process
determine consider enumerate define define define create
scope reuse terms classes properties constraints instances
In reality - an iterative process:
determine consider enumerate consider define enumerate define
scope reuse terms reuse classes terms classes
define define define define create define create
properties classes properties constraints instances classes instances
consider define define create
reuse properties constraints instances
Rules for FAIR vocabularies (deployment)
1. Provide a license that allows repurposing
2. Determine the content governance arrangements
3. Check minimal term definition completeness
4. Select a domain and service for the web identifiers, i.e. a namespace
5. Design an identifier schema and pattern
6. Create a semantic-standards based vocabulary – Interoperability
7. Add rich metadata – Reusability
8. Register the vocabulary, e.g. with LOV – Findability
9. Make the IRIs resolve – Accessibility
10. Implement a process for maintaining the vocabulary
Semantic-standards compliant vocabulary
1. Identify terms
2. Encode term labels and synonyms
3. Add textual definitions
4. Add notes or comments for clarifications
5. Add codes and symbols
6. Define the hierarchy of terms
7. Encode relationships
8. Define subsets
9. Define and document the whole vocabulary
Next
Advanced
OWL
