introduction to the semantic web - description...

29
Description Logic Basics Beyond ALC ITS 489 ICT SIIT Introduction to the Semantic Web Description Logics Boontawee Suntisrivaraporn [email protected] August 17, 2009 B Suntisrivaraporn Introduction to the Semantic Web 1

Upload: others

Post on 24-Jul-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

ITS 489 ◦ ICT ◦ SIIT

Introduction to the Semantic WebDescription Logics

Boontawee [email protected]

August 17, 2009

B Suntisrivaraporn ◦ Introduction to the Semantic Web 1

Page 2: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Outline

1 Description Logic Basics

2 Beyond ALC

B Suntisrivaraporn ◦ Introduction to the Semantic Web 2

Page 3: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Syntax...of the Description Logic ALC

Definition

Let CN and RN be two disjoint sets of concept and role names,respectively. Then the following are concept descriptions:

>, ⊥, A,

¬C ,

C u D,

C t D,

∃r .C ,

∀r .C ,

where A ∈ CN, r ∈ RN, and C , D are concept descriptions.

B Suntisrivaraporn ◦ Introduction to the Semantic Web 4

Page 4: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Examples...of ALC concept descriptions

Example

1 Person uMaleMale persons

2 Person u ∃child.>Persons that have a child

3 Father tMotherCollection of fathers and mothers

4 Mother u ∀child.HappyMothers whose all children are happy

5 Woman u ∀child.HappyWomen whose all children, if any, are happy

B Suntisrivaraporn ◦ Introduction to the Semantic Web 5

Page 5: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Semantics...of the Description Logic ALC

The semantics in DL is defined by means of interpretations,which is based on the set theory

Definition

An interpretation I is a pair (∆I , ·I) of

the interpretation domain ∆I and

the interpretation function ·IMapping each concept to a subset of ∆I

Mapping each role to a subset of ∆I ×∆I

B Suntisrivaraporn ◦ Introduction to the Semantic Web 6

Page 6: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Interpretations...of ALC concept descriptions

Definition

>I := ∆I

⊥I := ∅AI ⊆ ∆I

(¬C )I := ∆I\C I(C u D)I := C I ∩ DI

(C t D)I := C I ∪ DI

(∃r .C )I := {d | there exists e s.t. (d , e) ∈ rI and e ∈ C I}(∀r .C )I := {d | for all e, if (d , e) ∈ rI then e ∈ C I}

B Suntisrivaraporn ◦ Introduction to the Semantic Web 7

Page 7: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Examples...of an interpretation

Example

Let ∆I be {d0, d1, d2, d3, d4, d5},and ·I defined as follows:

PersonI ={d0, d1, d2, d3, d4}

MaleI = {d1, d3}

FemaleI = {d0, d2, d4}

HappyI = {d3, d4, d5}

1 (Person uMale)I

=

2 (Person u ∃child.>)I

=

3 (Father tMother)I

=

4 (Mother u ∀child.Happy)I

=

5 (Woman u ∀child.Happy)I

=

B Suntisrivaraporn ◦ Introduction to the Semantic Web 8

Page 8: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Concept AxiomsConcept definitions

TBox defines the terminology by means of concept definitions

Concept definition

A ≡ C(Complete) concept definitionConcept name A is defined to be precisely the same asconcept description C , specifying both necessary andsufficient conditions

A v CPrimitive concept definitionConcept name A is defined to satisfy the necessaryconditions given in concept description C

B Suntisrivaraporn ◦ Introduction to the Semantic Web 9

Page 9: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Examples...of ALC concept definitions

Example

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.PersonFather ≡ Man u ∃child.Person

Mother ≡ Woman u ∃child.PersonFatherWODaughter ≡ Father u ∀child.Male

ManWODaughter ≡ Man u ∀child.MaleFamily v SocialUnit u ∃part.Person

B Suntisrivaraporn ◦ Introduction to the Semantic Web 10

Page 10: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Concept AxiomsConcept definitions

The semantics is again based on interpretations

Semantics of concept definition

Let I be an interpretation.

I satisfies A ≡ C if AI = C I

I.e. A and C are interpreted to be the same set of objects

I satisfies A v C if AI ⊆ C I

I.e. A is interpreted to be a subset of C ’s interpretation

B Suntisrivaraporn ◦ Introduction to the Semantic Web 11

Page 11: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Interpretations...of ALC concept descriptions

Example

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.PersonFather ≡ Man u ∃child.Person

Mother ≡ Woman u ∃child.PersonFatherWODaughter ≡ Father u ∀child.Male

ManWODaughter ≡ Man u ∀child.MaleFatherWOChild ≡ Father u ∀child.⊥

B Suntisrivaraporn ◦ Introduction to the Semantic Web 12

Page 12: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Concept AxiomsGeneral concept inclusions

TBox can constrain on concepts readily defined somewhere

General concept inclusion (GCI)

C v DConcept description C is more specific than (or, mustimply) concept description D

Example

Man uMarried u ∃hasPet.Dog v HappyParent v ∃hasResponsibility.ChildRaise

B Suntisrivaraporn ◦ Introduction to the Semantic Web 13

Page 13: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Description Logic-based OntologyTBox & ABox

Definition

An ontology based on Description Logic, or knowledge based,comprises:

TBox—terminological partBackground knowledge about the domain in general

ABox—assertional partFacts about individual objects from the domain

B Suntisrivaraporn ◦ Introduction to the Semantic Web 14

Page 14: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

TBoxesVarious kinds of TBoxes abound

Unfoldable TBox

Only concept definitions can be used, and the followingconditions must be respected:

Definitional—there can be at most one definition foreach concept nameI.e. Impossible to have A ≡ C and A v D in TAcyclic—there is no cyclic dependency in TI.e. Impossible to have A ≡ B u C and B ≡ ∃r .A in T

General TBox

Both GCIs and concept definitions can be used!

B Suntisrivaraporn ◦ Introduction to the Semantic Web 15

Page 15: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Examples of TBoxes

Is this an unfoldable TBox?

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.Person

Is this an unfoldable TBox?

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.PersonPerson ≡ Man tWoman

B Suntisrivaraporn ◦ Introduction to the Semantic Web 16

Page 16: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

General TBox vs Unfoldable TBox

Every unfoldable TBox is a general TBox

Using GCIs to simulate concept definitions

A v C is a special kind of GCI

A ≡ C can be simulated by two GCIs to formulateimplications in both directions

A v C — the necessary conditionC v A — the sufficient condition

Is this a general TBox?

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.PersonPerson ≡ Man tWoman

B Suntisrivaraporn ◦ Introduction to the Semantic Web 17

Page 17: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Semantics of GCIs and TBoxes

The semantics is again based on interpretations

Semantics of GCIs

Let I be an interpretation. Then I satisfies C v D ifC I ⊆ DI

I.e. C is interpreted to be a subset of D’s interpretation

Semantics of TBox

Let T be a (general or unfoldable) TBox. An interpretation Iis a model of T if I satisfies all axioms in T

If A ≡ D, then AI = DI

If C v D, then C I ⊆ DI

B Suntisrivaraporn ◦ Introduction to the Semantic Web 18

Page 18: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Reasoning ProblemsConcept satisfiability

Concepts are intended to represent classes of objects

They should not be interpreted as the empty set ∅

Concept satisfiability

Concept C is satisfiable if there is an interpretation I suchthat CI 6= ∅

Concept C is satisfiable w.r.t. TBox T if there is a model Iof T such that CI 6= ∅

B Suntisrivaraporn ◦ Introduction to the Semantic Web 19

Page 19: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

ExamplesConcept satisfiability

This is an unsatisfiable concept

(Man u ∃child.Female) u ∀child.⊥

Are these satisfiable?

⊥, Female uMale, A t ¬A, A u ¬A, ∀r .⊥, ∃r .⊥

Now consider satisfiability w.r.t. TBox

Man ≡ Person uMaleWoman ≡ Person u Female

Male ≡ ¬Female

Is Female uMale satisfiable w.r.t. the TBox?

B Suntisrivaraporn ◦ Introduction to the Semantic Web 20

Page 20: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Reasoning ProblemsConcept subsumption

Concepts in DL correspond to classes in OWL

Concept subsumption in DL correspond to subclassrelation in OWL

Concept subsumption

Concept D subsumes C (C v D) if there, for allinterpretations I, CI ⊆ DI

Concept D subsumes C w.r.t. TBox T (C vT D) if there, forall models I of T , CI ⊆ DI

B Suntisrivaraporn ◦ Introduction to the Semantic Web 21

Page 21: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

ExamplesConcept subsumption

Without TBox

∃child.Doctor t ∃child.Lawyer v ∃child.(Doctor t Lawyer)∀child.(Rich u Happy) v ∀child.Rich u ∀child.Happy

Which subsumptions hold true?

⊥ v A, A v >, A u B v A, A v A t B

Now consider subsumption w.r.t. TBox

Parent ≡ Person u ∃child.PersonFather ≡ Man u ∃child.Person

Man ≡ Person uMale

Is it the case that Father vT Parent?

B Suntisrivaraporn ◦ Introduction to the Semantic Web 22

Page 22: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Syntax...of the Description Logic ALC

Definition

Let CN and RN be two disjoint sets of concept and role names,respectively. Then the following are concept descriptions:

Atomic concepts >, ⊥, A,

Negation ¬C ,

Conjunction C u D,

Disjunction C t D,

Existential restriction ∃r .C ,

Value restriction ∀r .C ,

where A ∈ CN, r ∈ RN, and C , D are concept descriptions.

B Suntisrivaraporn ◦ Introduction to the Semantic Web 24

Page 23: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Syntax...of the Description Logic SHOIN & SHOIQ

Definition

Apart from CN and RN, we have a third disjoint set ofindividuals Ind. We can use all the concept constructs inALC, and MORE!

Nominal concept {a}Unqualified number restriction (≤ n r), (≥ n r)

Qualified number restriction (≤ n r C ), (≥ n r C )

where a ∈ Ind, r ∈ RN, and C is concept description.

SHOIQ is more than just a concept language

S stands for ALC plus role transitivity: trans(r)

H stands for role hierarchy: r v s

I stands for inverse role: r−B Suntisrivaraporn ◦ Introduction to the Semantic Web 25

Page 24: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Semantics...of Description Logics

The semantics in DL is defined by means of interpretations,which is based on the set theory

Definition

An interpretation I is a pair (∆I , ·I) of

the interpretation domain ∆I and

the interpretation function ·IMapping each concept to a subset of ∆I

Mapping each role to a subset of ∆I ×∆I

Mapping each individual a to aI ∈ ∆I

B Suntisrivaraporn ◦ Introduction to the Semantic Web 26

Page 25: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Interpretations...of ALC concept descriptions

Definition

>I := ∆I

⊥I := ∅AI ⊆ ∆I

(¬C )I := ∆I\C I(C u D)I := C I ∩ DI

(C t D)I := C I ∪ DI

(∃r .C )I := {d | there exists e s.t. (d , e) ∈ rI and e ∈ C I}(∀r .C )I := {d | for all e, if (d , e) ∈ rI then e ∈ C I}

B Suntisrivaraporn ◦ Introduction to the Semantic Web 27

Page 26: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Interpretations...of additional constructors in SHOIN & SHOIQ

Concept constructors

{a}I := {aI}(≤ n r)I := {d | ]{e : (d , e) ∈ rI} ≤ n}(≥ n r)I := {d | ]{e : (d , e) ∈ rI} ≥ n}

(≤ n r C )I := {d | ]{e : (d , e) ∈ rI and e ∈ C I} ≤ n}(≥ n r C )I := {d | ]{e : (d , e) ∈ rI and e ∈ C I} ≥ n}

B Suntisrivaraporn ◦ Introduction to the Semantic Web 28

Page 27: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Interpretations...of additional constructors in SHOIN & SHOIQ

Role constructors

(r−)I := {(b, a) | (a, b) ∈ rI}

Role axiom constructors

Interpretation I satisfies

trans(r)if, for all d , e, f ∈ ∆I , (d , e), (e, f ) ∈ rI ⇒ (d , f ) ∈ rI

r v sif rI ⊆ sI

B Suntisrivaraporn ◦ Introduction to the Semantic Web 29

Page 28: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Examples...of SHOIQ concept definitions

Example

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.PersonPerson v ∃child−.Person

Chinese v Person u (≤ 1 child Chinese)

B Suntisrivaraporn ◦ Introduction to the Semantic Web 30

Page 29: Introduction to the Semantic Web - Description Logicsict.siit.tu.ac.th/~sun/SW/its489-dl-handout.pdf · B Suntisrivaraporn Introduction to the Semantic Web 10. Description Logic BasicsBeyond

Description Logic Basics Beyond ALC

Examples...of SHOIQ concept definitions

Example

Man ≡ Person uMaleWoman ≡ Person u Female

Parent ≡ Person u ∃child.PersonPerson v ∃child−.Person

Chinese v Person u (≤ 1 child Chinese)ParentA ≡ Parent u ∃child.Rich u ∃child.HappyParentB ≡ Parent u ∃child.Male u ∃child.FemaleParentC ≡ Parent u (≥ 2 child)

B Suntisrivaraporn ◦ Introduction to the Semantic Web 31