OWL verses UML Jan Pettersen Nytun, UiA 1 S O P [1]: The - - PowerPoint PPT Presentation

owl verses uml
SMART_READER_LITE
LIVE PREVIEW

OWL verses UML Jan Pettersen Nytun, UiA 1 S O P [1]: The - - PowerPoint PPT Presentation

Knowledge Representation Part VIIc OWL verses UML Jan Pettersen Nytun, UiA 1 S O P [1]: The languages were devised to fulfill different purposes. While OWL supports the representation of knowledge about a system, UML was developed


slide-1
SLIDE 1

1

Knowledge Representation Part VIIc

OWL verses UML

Jan Pettersen Nytun, UiA

slide-2
SLIDE 2

S O

P

[1]: “The languages were devised to fulfill different purposes. While OWL supports the representation of knowledge about a system, UML was developed primarily to support the construction of a (software) system.”

Jan Pettersen Nytun, UiA, page 2

slide-3
SLIDE 3

S O

P

Open-World versus Closed-World Interpretation Assumptions

[1]: “The UML is oriented towards data modeling and system construction … knowledge is implicitly viewed as being complete. OWL, in contrast, interprets models as potentially representing partial knowledge.”

Jan Pettersen Nytun, UiA, page 3

slide-4
SLIDE 4

S O

P

[1] Unique Name Assumption & Synonyms

  • OWL allows the definition of synonyms for

classes, properties and individual descriptions.

  • UML assumes that every name has a unique

interpretation, i.e., the unique name assumption.

Jan Pettersen Nytun, UiA, page 4

slide-5
SLIDE 5

S O

P

[1] Sufficient Conditions & Defined vs. Primitive Concepts

  • One of the purposes of OWL ontologies is

to facilitate automatic classification.

  • The UML does not provide native

assistance in the definition of sufficient conditions or defined classes.

Jan Pettersen Nytun, UiA, page 5

slide-6
SLIDE 6

S O

P

UML  OWL

Jan Pettersen Nytun, UiA, page 6

Person name : String Institution name : String

employee employedAt

* 1

slide-7
SLIDE 7

S O

P

Jan Pettersen Nytun, UiA, page 7

Person name : String In UML a class is a namespace by itself. Institution name : String I.e., property name in class Person and property name in class Institution are different entities. Properties in OWL are globally scoped. Why is the following not a good idea?

slide-8
SLIDE 8

S O

P

Simulate the namespace of the class by using class name as prefix.

Jan Pettersen Nytun, UiA, page 8

Person hasPersonName : String

slide-9
SLIDE 9

S O

P

Multiplicity of hasPersonName is by default one.

Jan Pettersen Nytun, UiA, page 9

Person hasPersonName : String

:Person rdf:type owl:Class ; rdfs:subClassOf [ rdf:type owl:Restriction ;

  • wl:onProperty :hasPersonName ;
  • wl:qualifiedCardinality

"1"^^xsd:nonNegativeInteger ;

  • wl:onDataRange xsd:string

] .

slide-10
SLIDE 10

S O

P

Starting the reasoner – what is inferred?

Jan Pettersen Nytun, UiA, page 10

slide-11
SLIDE 11

S O

P

Jan Pettersen Nytun, UiA, page 11

slide-12
SLIDE 12

S O

P

Jan Pettersen Nytun, UiA, page 12

Remove Person as Domain and run reasoner – what is inferred?

slide-13
SLIDE 13

S O

P

Jan Pettersen Nytun, UiA, page 13

Nothing is inferred!

slide-14
SLIDE 14

S O

P

Jan Pettersen Nytun, UiA, page 14

Make Person a defined class. Run reasoner – what is inferred?

slide-15
SLIDE 15

S O

P

We have an individual with one hasPersonName property specified. But still this individual is not classified as a Person - why?

Jan Pettersen Nytun, UiA, page 15

slide-16
SLIDE 16

S O

P

  • OWL makes the open world assumption.
  • In our example this means: There may be

more hasPersonName properties for our individual.

Jan Pettersen Nytun, UiA, page 16

slide-17
SLIDE 17

S O

P

  • It is possible to specify that the individual has
  • nly "Jan Pettersen"^^xsd:string

as value for property hasPersonName.

Jan Pettersen Nytun, UiA, page 17

Inferred by reasoner

slide-18
SLIDE 18

S O

P

Better solution:

Jan Pettersen Nytun, UiA, Ontologies, page 18

slide-19
SLIDE 19

S O

P

UML  OWL

Jan Pettersen Nytun, UiA, page 19

Person name : String Institution name : String

employee employedAt

* 1

Person name : String Institution name : String

institutionHas- EmployedPerson personIsEmployed- AtInstitution

* 1

“unique naming”

slide-20
SLIDE 20

S O

P

Jan Pettersen Nytun, UiA, page 20

slide-21
SLIDE 21

S O

P

Jan Pettersen Nytun, UiA, Ontologies, page 21

slide-22
SLIDE 22

S O

P

References

Jan Pettersen Nytun, UiA, Propositional Calculus, page 22

[1] A Detailed Comparison of UML and OWL, Kilian Kiko & Colin Atkinson, https://ub-madoc.bib.uni-mannheim.de/1898/1/TR2008_004.pdf