Ontology Development

1
Ontology Development

• Kenneth Baclawski
• Northeastern University
• Harvard Medical School

2
Outline

• Background on ontologies
• Ontology development phases
• Requirements and Analysis
• Design and Implementation
• Testing and Validation
• Maintenance
• Comparison of ontologies and software
• Classes
• Relationships
• Logic
• Conclusion

3
Ontologies

• Ontology What exists in a domain and how they
  relate with each other.
• Formal ontology Formal treatment of the concepts
  and relationships in a domain.
• Simple Example
• Employees work for Companies
• Employees report to Employees

4
Statements
An object is an instance of the Employee class.
George is an employee.
An object in the Employee class is linked with
an object in the Company class via the works_for
relationship.
George works for Sony.
An object in the Employee class is linked with
another object in the same class via the
reports_to relationship.
George reports to Adam.
Fred works for a company.
George says that Fred works for Toyota.
Fred reports to two other employees.
Fred must report to a another employee.
5
Purposes of Ontologies

• Basis for communication
• Between people (may be informal)
• Between agents (formal ontologies)
• Applications
• Representing and storing data (e.g., DB schema)
• Knowledge sharing within and between domains
• Search and retrieval
• Software development
• Classification and organization of data resources
• Establishing contracts
• Policy enforcement

6
Criteria for Introducing Ontologies

• Large amounts of data
• Data available on the Web
• Data acquired or generated by new techniques
• Complex data structures
• Inheritance, containment and other hierarchies
• Many relationships
• Diverse sources
• Many legacy systems
• Sources on the Web using different formats
• Requirement for formal proofs
• Contracts and policy enforcement

7
Ontology Development Today

• The ideal is for the tools to be based on
  methodologies and processes.
• The reality is the reverse methodologies are
  based on the tools.
• Opportunity The lack of good ontology
  development processes and methodologies
• Thesis Given that formal ontologies are a form
  of software, software development methodologies
  can be adapted to serve ontology development.

8
Classification of Ontology Languages

• Logical languages
• First order predicate logic
• Rule based logic
• Description logic
• Frame based languages
• Similar to relational databases
• Graph based languages
• Semantic networks
• Analogy with the Web is rationale for the
  Semantic Web

9
Some Ontology Languages

• Established languages
• Knowledge Interchange Format (KIF)
• XML Schema (XSD)
• Resource Description Framework (RDF)
• XML Topic Maps (XTM)
• Emerging languages
• Common Logic
• Web Ontology Languages (OWL)
• Ontology Definition Metamodel (ODM)

10
Ontology Development Phases

• Requirements and Analysis
• Design and Implementation
• Testing and Validation
• Maintenance

11
Requirements and Analysis

• Least understood of all phases
• Direct involvement by stakeholders is essential,
  but how?
• Specifying the scope is important, but not all
  languages support it.
• Point of view is also relevant.
• These phases offer significant opportunities for
  new methodologies and processes.

12
Design and Implementation Patterns and Reuse

• Design pattern Use known patterns
• Metaphor Transform another ontology
• Composition Transform and combine many smaller
  ontologies

13
Example of ontology composition
Fuzzy Geographic Ontology
Geographic Ontology
Fuzzy Logic
Boolean Logic

• The arrows are theory morphisms in the category
  of theories.
• The composition is the colimit the theories are
  combined such that the common features are
  equivalent.
• The composition not only allows one to express
  fuzzy geographic statements but also to propagate
  uncertainty from base facts (observations) to
  derived facts via rules.

14
Design and Implementation Refactoring

• Move methods from one class to another (for
  example, from subclasses to superclasses).
• Reification and unreification changing
  relationships into classes or vice versa.
• Metalevel shifting (for example, reflection)
• Refactoring adjusts ontological commitment.
• Refactoring is useful at the design,
  implementation and maintenance phases.

15
Example of Refactoring
16
Example of Reification
17
Testing and Validation

• Validation of requirements

• Consistency checking

18
Comparison of ontologies and software

• Classes
• Sets versus templates
• Behavior versus set-theory
• Properties
• Aspects
• Logic
• Open versus closed

19
Class Set or Template

• Object-oriented classes are templates. They
  allow one to construct objects with specified
  features.
• Formal ontological classes are sets which can be
  defined in terms of other classes, attributes and
  associations.

20
Derived class example

• When the age of a person is updated, membership
  in the Teenager class may also be changed.
• When testing for membership in a class, one
  should use isCompatibleWith or isConsistentWith
  rather than instanceof.

21
Class Set-theory or Behavior
22
Properties

• Property is a general term for attributes and
  associations.
• Are properties first class or second class?
• First class properties are an example of an
  aspect a feature that cuts across class
  boundaries.
• First class properties are part of UML2.0.

23
Logic Open versus Closed

• Open (monotonic) logic gives different answers to
  queries than a closed logic.

• Suppose that Fred is just an employee
• Closed world violates constraint
• Open world Fred works for a company, but the
  company is not known.

24
Logic Open versus Closed

• Suppose that Fred works for both Sony and IBM
• Closed world violates constraint
• Open world Sony and IBM are the same company!

25
Statements

• Fred works for a company.
• The Fred object is an instance of the class of
  employees who work for at least one company.
• Fred reports to two other employees.
• The Fred object is an instance of the class of
  employees who report to exactly two employees.
• The Fred object is an instance of the class of
  employees who do not report to themselves.

26
Statements

• Fred must report to a another employee.
• The Fred object is an instance of the class of
  employees who report to at least one employee.
• The Fred object is an instance of the class of
  employees who do not report to themselves.
• George says that Fred works for Toyota.
• Homework assignment.
• Hint Use reification.

27
Conclusion

• There is a noteworthy lack of methodologies and
  process models for ontology development.
• Software development methodologies and processes
  could be adapted for ontologies.
• However, many challenges remain to be solved.

28
Some of my efforts toward adapting software
development for ontologies

• Extend UML to support ontologies
• The UML has approved some of my suggestions
• Direct support for ontology development by CASE
  tools
• The OMG has issued an RFP
• Four initial submissions have been presented.
• Use cases are being developed
• Transformations between formal languages
• The OMG has issued an RFP
• Six submissions are at the revised submission
  stage.
• Several products already exist