An Introduction to Description Logicsesslli2018.folli.info/wp-content/uploads/ESSLLI-Part2.pdfIvan...
Transcript of An Introduction to Description Logicsesslli2018.folli.info/wp-content/uploads/ESSLLI-Part2.pdfIvan...
-
Making Statements DL Knowledge Bases Entailment in DLs
An Introduction to Description LogicsPart 2: Formal Ontologies in ALC
Ivan Varzinczak
CRIL, Univ. Artois & CNRSLens, France
http://www.ijv.ovh
ESSLLI 2018
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 1
http://www.ijv.ovh
-
Making Statements DL Knowledge Bases Entailment in DLs
Recap: Main ingredients in formal ontologiesA common vocabulary and a shared understanding
Classes or concepts• Describe concrete or abstract entities within the domain of interest
• E.g.: Employed student, Parent
Relations or roles• Describe relationships between concepts or attributes of a concept
• E.g.: work for someone, being employed by someone
Instances of classes and relations• Name objects of the domain and denote representatives of a concept
• E.g.: John, John is an employed student, John works for IBM
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 2
-
Making Statements DL Knowledge Bases Entailment in DLs
Recap: Why Description Logics?Expressivity• Concepts X
• Relations X
• Instances X
DLs have all one needs to formalise ontologies!
Available tools
Computational properties• Amenability to implementation X
• Decidability X
• Good trade-off between expressivity and complexity X
Most DL-based systems satisfy all of these!
FaCT++
Pellet
HermiT
CEL
· · ·
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 3
-
Making Statements DL Knowledge Bases Entailment in DLs
Recap: Concept languageAtomic concept names• C =def {A1, . . . , An} (Special concepts: >, ⊥)• Intuition: basic classes of a domain of interest• Student, Employee, Parent
Atomic role names• R =def {r1, . . . , rm}• Intuition: basic relations between concepts• worksFor, empBy
Individual names• I =def {a1, . . . , al}• Intuition: names of objects in the domain• john, mary, ibm
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 4
-
Making Statements DL Knowledge Bases Entailment in DLs
Recap: Concept languageBoolean constructors• Concept negation: ¬ (class complement)• Concept conjunction: u (class intersection)• Concept disjunction: t (class union)Role restrictions
• Existential restriction: ∃ (at least one relationship)• Value restriction: ∀ (all relationships)Further constructors: cardinality constraints, inverse roles, . . . (if needed)
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 5
-
Making Statements DL Knowledge Bases Entailment in DLs
Recap: Concept language
ALC complex concepts: LALC
>, ⊥ (constants)
A (atomic concept)
¬C (complement)
C uD (conjunction)
C tD (disjunction)
∃r.C (existential restriction)
∀r.C (value restriction)
PlusIndividual names a
Semantics: I = 〈∆I , ·I〉
>I = ∆I , ⊥I = ∅
AI ⊆ ∆I
∆I \ CI
CI ∩DI
CI ∪DI
{x ∈ ∆I | rI(x) ∩ CI 6= ∅}
{x ∈ ∆I | rI(x) ⊆ CI}
AndaI ∈ ∆I
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 6
-
Making Statements DL Knowledge Bases Entailment in DLs
Outline
Making Statements
DL Knowledge Bases
Entailment in DLs
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 7
-
Making Statements DL Knowledge Bases Entailment in DLs
Outline
Making Statements
DL Knowledge Bases
Entailment in DLs
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 8
-
Making Statements DL Knowledge Bases Entailment in DLs
MotivationConcept language of ALC
>, ⊥ (constants)A (atomic concept)¬C (complement of C)C uD (intersection of C and D)C tD (union of C and D)∃r.C (existential restriction)∀r.C (value restriction)
Something is missing
• The central notion in logic: C ‘→’ D• What would C ‘→’ D mean here?• DLs have a version of ‘→’ that is very special
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 9
-
Making Statements DL Knowledge Bases Entailment in DLs
MotivationConcept language of ALC
>, ⊥ (constants)A (atomic concept)¬C (complement of C)C uD (intersection of C and D)C tD (union of C and D)∃r.C (existential restriction)∀r.C (value restriction)
Something is missing
• The central notion in logic: C ‘→’ D• What would C ‘→’ D mean here?• DLs have a version of ‘→’ that is very special
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 9
-
Making Statements DL Knowledge Bases Entailment in DLs
MotivationConcept language of ALC
>, ⊥ (constants)A (atomic concept)¬C (complement of C)C uD (intersection of C and D)C tD (union of C and D)∃r.C (existential restriction)∀r.C (value restriction)
Something is missing• The central notion in logic: C ‘→’ D
• What would C ‘→’ D mean here?• DLs have a version of ‘→’ that is very special
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 9
-
Making Statements DL Knowledge Bases Entailment in DLs
MotivationConcept language of ALC
>, ⊥ (constants)A (atomic concept)¬C (complement of C)C uD (intersection of C and D)C tD (union of C and D)∃r.C (existential restriction)∀r.C (value restriction)
Something is missing• The central notion in logic: C ‘→’ D• What would C ‘→’ D mean here?
• DLs have a version of ‘→’ that is very special
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 9
-
Making Statements DL Knowledge Bases Entailment in DLs
MotivationConcept language of ALC
>, ⊥ (constants)A (atomic concept)¬C (complement of C)C uD (intersection of C and D)C tD (union of C and D)∃r.C (existential restriction)∀r.C (value restriction)
Something is missing• The central notion in logic: C ‘→’ D• What would C ‘→’ D mean here? (We already have ¬C tD)
• DLs have a version of ‘→’ that is very special
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 9
-
Making Statements DL Knowledge Bases Entailment in DLs
MotivationConcept language of ALC
>, ⊥ (constants)A (atomic concept)¬C (complement of C)C uD (intersection of C and D)C tD (union of C and D)∃r.C (existential restriction)∀r.C (value restriction)
Something is missing• The central notion in logic: C ‘→’ D• What would C ‘→’ D mean here? (We already have ¬C tD)• DLs have a version of ‘→’ that is very special
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 9
-
Making Statements DL Knowledge Bases Entailment in DLs
Statements
In many logics
meta-language(entailment, etc)
object language(formulae)
In DLs
meta-language
statements
concept language
• Two levels of language• Two notions of ‘entailment’• Two notions of ‘satisfaction’
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 10
-
Making Statements DL Knowledge Bases Entailment in DLs
Statements
In many logics
meta-language(entailment, etc)
object language(formulae)
In DLs
meta-language
statements
concept language
• Two levels of language• Two notions of ‘entailment’• Two notions of ‘satisfaction’
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 10
-
Making Statements DL Knowledge Bases Entailment in DLs
Making statementsSubsumption• Concept inclusion
• Employed students are students
• Employed students are employees
Instantiation or assertions• Concept and role membership
• John is an employed student (John instantiates employed student)
• John works for IBM (John and IBM instantiate works for)
Statements talk about concepts, roles and individualsThey are not concepts! They are in the ‘in-between’ language
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 11
-
Making Statements DL Knowledge Bases Entailment in DLs
Making statementsSubsumption• Concept inclusion
• Employed students are students
• Employed students are employees
Instantiation or assertions• Concept and role membership
• John is an employed student (John instantiates employed student)
• John works for IBM (John and IBM instantiate works for)
Statements talk about concepts, roles and individualsThey are not concepts! They are in the ‘in-between’ language
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 11
-
Making Statements DL Knowledge Bases Entailment in DLs
Making statementsSubsumption• Concept inclusion
• Employed students are students
• Employed students are employees
Instantiation or assertions• Concept and role membership
• John is an employed student (John instantiates employed student)
• John works for IBM (John and IBM instantiate works for)
Statements talk about concepts, roles and individuals
They are not concepts! They are in the ‘in-between’ language
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 11
-
Making Statements DL Knowledge Bases Entailment in DLs
Making statementsSubsumption• Concept inclusion
• Employed students are students
• Employed students are employees
Instantiation or assertions• Concept and role membership
• John is an employed student (John instantiates employed student)
• John works for IBM (John and IBM instantiate works for)
Statements talk about concepts, roles and individualsThey are not concepts! They are in the ‘in-between’ language
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 11
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C v DIntuition• D subsumes C (or C is subsumed by D)
• C is more specific than D (or D is more general than C)
• Formalise one aspect of is-a relations
Example• EmpStud v Student u Employee, Employee v ∃worksFor.>
• EmpStud v ∃pays.Tax, Student u ¬Employee v ¬∃pays.Tax
Central notion in DL terminologies (taxonomies)
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 12
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C v DIntuition• D subsumes C (or C is subsumed by D)
• C is more specific than D (or D is more general than C)
• Formalise one aspect of is-a relations
Example• EmpStud v Student u Employee, Employee v ∃worksFor.>
• EmpStud v ∃pays.Tax, Student u ¬Employee v ¬∃pays.Tax
Central notion in DL terminologies (taxonomies)
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 12
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C v DIntuition• D subsumes C (or C is subsumed by D)
• C is more specific than D (or D is more general than C)
• Formalise one aspect of is-a relations
Example• EmpStud v Student u Employee, Employee v ∃worksFor.>
• EmpStud v ∃pays.Tax, Student u ¬Employee v ¬∃pays.Tax
Central notion in DL terminologies (taxonomies)
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 12
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C v DSemantics• I C v D (I satisfies C v D) if CI ⊆ DI
• First level of ‘entailment’: all C-objects are D-objects
∆I
CI
DI
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 13
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C v DSemantics• I C v D (I satisfies C v D) if CI ⊆ DI
• First level of ‘entailment’: all C-objects are D-objects
∆I
CI
DI
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 13
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C ≡ DConcept equivalence• Just an abbreviation for C v D and D v C• I C ≡ D if I C v D and I D v C• I C ≡ D if CI = DI
∆I
CI DI
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 14
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumption statements
C ≡ DConcept equivalence• Just an abbreviation for C v D and D v C• I C ≡ D if I C v D and I D v C• I C ≡ D if CI = DI
∆I
CI DI
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 14
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee• I ∃worksFor.> v Employee• I Employee v ∃worksFor.>• I EmpStud v ∀pays.Tax
• I Parent u ¬Employee v Tax t ¬Student• I EmpStud u Parent v ∃pays.Tax• I ∃empBy.> v ∃worksFor.Company• I ∀pays.Tax v ∃empBy.Company
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee ?• I ∃worksFor.> v Employee ?• I Employee v ∃worksFor.> ?• I EmpStud v ∀pays.Tax ?
• I Parent u ¬Employee v Tax t ¬Student ?• I EmpStud u Parent v ∃pays.Tax ?• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee ?• I Employee v ∃worksFor.> ?• I EmpStud v ∀pays.Tax ?
• I Parent u ¬Employee v Tax t ¬Student ?• I EmpStud u Parent v ∃pays.Tax ?• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> ?• I EmpStud v ∀pays.Tax ?
• I Parent u ¬Employee v Tax t ¬Student ?• I EmpStud u Parent v ∃pays.Tax ?• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> No!• I EmpStud v ∀pays.Tax ?
• I Parent u ¬Employee v Tax t ¬Student ?• I EmpStud u Parent v ∃pays.Tax ?• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> No!• I EmpStud v ∀pays.Tax No!
• I Parent u ¬Employee v Tax t ¬Student ?• I EmpStud u Parent v ∃pays.Tax ?• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> No!• I EmpStud v ∀pays.Tax No!
• I Parent u ¬Employee v Tax t ¬Student Yep!• I EmpStud u Parent v ∃pays.Tax ?• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> No!• I EmpStud v ∀pays.Tax No!
• I Parent u ¬Employee v Tax t ¬Student Yep!• I EmpStud u Parent v ∃pays.Tax No!• I ∃empBy.> v ∃worksFor.Company ?• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> No!• I EmpStud v ∀pays.Tax No!
• I Parent u ¬Employee v Tax t ¬Student Yep!• I EmpStud u Parent v ∃pays.Tax No!• I ∃empBy.> v ∃worksFor.Company Yep!• I ∀pays.Tax v ∃empBy.Company ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I EmpStud v Student u Employee Yep!• I ∃worksFor.> v Employee Yep!• I Employee v ∃worksFor.> No!• I EmpStud v ∀pays.Tax No!
• I Parent u ¬Employee v Tax t ¬Student Yep!• I EmpStud u Parent v ∃pays.Tax No!• I ∃empBy.> v ∃worksFor.Company Yep!• I ∀pays.Tax v ∃empBy.Company No!
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 15
-
Making Statements DL Knowledge Bases Entailment in DLs
Assertions
a : C (a, b) : rIntuition• a is an instance of C
• a and b are related via r (or (a, b) is an instance of r)
• Formalise another aspect of is-a relations
Example• john : EmpStud, mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax
• (john, ibm) : worksFor (in some DLs: (john, ibm) : ¬empBy)
Central notion in DL ‘databases’
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 16
-
Making Statements DL Knowledge Bases Entailment in DLs
Assertions
a : C (a, b) : rIntuition• a is an instance of C
• a and b are related via r (or (a, b) is an instance of r)
• Formalise another aspect of is-a relations
Example• john : EmpStud, mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax
• (john, ibm) : worksFor
(in some DLs: (john, ibm) : ¬empBy)
Central notion in DL ‘databases’
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 16
-
Making Statements DL Knowledge Bases Entailment in DLs
Assertions
a : C (a, b) : rIntuition• a is an instance of C
• a and b are related via r (or (a, b) is an instance of r)
• Formalise another aspect of is-a relations
Example• john : EmpStud, mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax
• (john, ibm) : worksFor (in some DLs: (john, ibm) : ¬empBy)
Central notion in DL ‘databases’
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 16
-
Making Statements DL Knowledge Bases Entailment in DLs
Assertions
a : C (a, b) : rIntuition• a is an instance of C
• a and b are related via r (or (a, b) is an instance of r)
• Formalise another aspect of is-a relations
Example• john : EmpStud, mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax
• (john, ibm) : worksFor (in some DLs: (john, ibm) : ¬empBy)
Central notion in DL ‘databases’
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 16
-
Making Statements DL Knowledge Bases Entailment in DLs
Assertions
a : C (a, b) : rSemantics• I a : C (I satisfies a : C) if aI ∈ CI
• I (a, b) : r (I satisfies (a, b) : r) if (aI , bI) ∈ rI
• First level of ‘satisfaction’: a is a ‘model’ of C, (a, b) is a ‘model’ of r
∆I
aI
CI
∆I
aI bI
rI
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 17
-
Making Statements DL Knowledge Bases Entailment in DLs
Assertions
a : C (a, b) : rSemantics• I a : C (I satisfies a : C) if aI ∈ CI
• I (a, b) : r (I satisfies (a, b) : r) if (aI , bI) ∈ rI
• First level of ‘satisfaction’: a is a ‘model’ of C, (a, b) is a ‘model’ of r
∆I
aI
CI
∆I
aI bI
rI
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 17
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax• I mary : ∀pays.Tax• I (mary, ibm) : empBy• I (ibm, john) : worksFor
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax• I mary : Employee u ∃empBy.>• I john : ∀empBy.Company• I john : ∃worksFor.∀pays.⊥
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax ?• I mary : ∀pays.Tax ?• I (mary, ibm) : empBy ?• I (ibm, john) : worksFor ?
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax ?• I mary : Employee u ∃empBy.> ?• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax ?• I (mary, ibm) : empBy ?• I (ibm, john) : worksFor ?
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax ?• I mary : Employee u ∃empBy.> ?• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy ?• I (ibm, john) : worksFor ?
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax ?• I mary : Employee u ∃empBy.> ?• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy No!• I (ibm, john) : worksFor ?
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax ?• I mary : Employee u ∃empBy.> ?• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy No!• I (ibm, john) : worksFor No!
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax ?• I mary : Employee u ∃empBy.> ?• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy No!• I (ibm, john) : worksFor No!
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax Yep!• I mary : Employee u ∃empBy.> ?• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy No!• I (ibm, john) : worksFor No!
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax Yep!• I mary : Employee u ∃empBy.> No!• I john : ∀empBy.Company ?• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy No!• I (ibm, john) : worksFor No!
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax Yep!• I mary : Employee u ∃empBy.> No!• I john : ∀empBy.Company Yep!• I john : ∃worksFor.∀pays.⊥ ?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
• I john : Employee u ∃pays.Tax Yep!• I mary : ∀pays.Tax Yep!• I (mary, ibm) : empBy No!• I (ibm, john) : worksFor No!
• I mary : Parent u ¬∃worksFor.> u ¬∃pays.Tax Yep!• I mary : Employee u ∃empBy.> No!• I john : ∀empBy.Company Yep!• I john : ∃worksFor.∀pays.⊥ Yep!
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 18
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumptions and assertionsValidity• Let α denote a statement
• |= α (α is valid) if I α for every I
Example• |= ¬(C uD) ≡ (¬C t ¬D)
• |= ∀r.(C uD) v ∀r.C
• 6|= ∃r.> v ∃r.C
• |= a : C t ¬C
• |= ¬(C tD) ≡ (¬C u ¬D)
• 6|= ∀r.C v ∀r.(C uD)
• |= ∃r.C v ∃r.>
• 6|= (a, b) : r
Watch out: Statements can be valid; concepts cannot!
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 19
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumptions and assertionsValidity• Let α denote a statement
• |= α (α is valid) if I α for every I
Example• |= ¬(C uD) ≡ (¬C t ¬D)
• |= ∀r.(C uD) v ∀r.C
• 6|= ∃r.> v ∃r.C
• |= a : C t ¬C
• |= ¬(C tD) ≡ (¬C u ¬D)
• 6|= ∀r.C v ∀r.(C uD)
• |= ∃r.C v ∃r.>
• 6|= (a, b) : r
Watch out: Statements can be valid; concepts cannot!
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 19
-
Making Statements DL Knowledge Bases Entailment in DLs
Subsumptions and assertionsValidity• Let α denote a statement
• |= α (α is valid) if I α for every I
Example• |= ¬(C uD) ≡ (¬C t ¬D)
• |= ∀r.(C uD) v ∀r.C
• 6|= ∃r.> v ∃r.C
• |= a : C t ¬C
• |= ¬(C tD) ≡ (¬C u ¬D)
• 6|= ∀r.C v ∀r.(C uD)
• |= ∃r.C v ∃r.>
• 6|= (a, b) : r
Watch out: Statements can be valid; concepts cannot!
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 19
-
Making Statements DL Knowledge Bases Entailment in DLs
Outline
Making Statements
DL Knowledge Bases
Entailment in DLs
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 20
-
Making Statements DL Knowledge Bases Entailment in DLs
TBoxes and ABoxesIntensional knowledge• Set of subsumption statements• Intuition: provide definitions of concepts (a terminology)• Called the TBox (terminological box). Notation: T
Extensional knowledge• Concept and role assertions• Intuition: provide an instantiation of concepts and roles (a ‘database’)• Called the ABox (assertion box). Notation: A
Definition (Knowledge base)A DL knowledge base (a.k.a. ontology) is a tuple KB =def 〈T ,A〉
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 21
-
Making Statements DL Knowledge Bases Entailment in DLs
TBoxes and ABoxesIntensional knowledge• Set of subsumption statements• Intuition: provide definitions of concepts (a terminology)• Called the TBox (terminological box). Notation: T
Extensional knowledge• Concept and role assertions• Intuition: provide an instantiation of concepts and roles (a ‘database’)• Called the ABox (assertion box). Notation: A
Definition (Knowledge base)A DL knowledge base (a.k.a. ontology) is a tuple KB =def 〈T ,A〉
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 21
-
Making Statements DL Knowledge Bases Entailment in DLs
TBoxes and ABoxesIntensional knowledge• Set of subsumption statements• Intuition: provide definitions of concepts (a terminology)• Called the TBox (terminological box). Notation: T
Extensional knowledge• Concept and role assertions• Intuition: provide an instantiation of concepts and roles (a ‘database’)• Called the ABox (assertion box). Notation: A
Definition (Knowledge base)A DL knowledge base (a.k.a. ontology) is a tuple KB =def 〈T ,A〉
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 21
-
Making Statements DL Knowledge Bases Entailment in DLs
Knowledge bases
Example (The student ontology in DL)
T =
EmpStud ≡ Student u Employee,Student u ¬Employee v ¬∃pays.Tax,
EmpStud u ¬Parent v ∃pays.Tax,EmpStud u Parent v ¬∃pays.Tax,∃worksFor.Company v Employee
A =
ibm : Company,mary : Parent,
john : EmpStud,(john, ibm) : worksFor
classesrelations
individuals
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 22
-
Making Statements DL Knowledge Bases Entailment in DLs
Knowledge basesSemantics• I T if I C v D for every C v D ∈ T• I A if:
• I a : C for every a : C ∈ A, and• I (a, b) : r for every (a, b) : r ∈ A
Moreover• If I T , we say I is a model of T• If I A, we say I is a model of A• If I T ∪ A, then I is a model of KB = 〈T ,A〉• KB is satisfiable if it has a model
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 23
-
Making Statements DL Knowledge Bases Entailment in DLs
Knowledge basesSemantics• I T if I C v D for every C v D ∈ T• I A if:
• I a : C for every a : C ∈ A, and• I (a, b) : r for every (a, b) : r ∈ A
Moreover• If I T , we say I is a model of T• If I A, we say I is a model of A• If I T ∪ A, then I is a model of KB = 〈T ,A〉• KB is satisfiable if it has a model
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 23
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise• Let KB = 〈T ,A〉, where:
T =
EmpStud ≡ Student u Employee,Student u ¬Employee v ¬∃pays.Tax,
EmpStud u ¬Parent v ∃pays.Tax,EmpStud u Parent v ¬∃pays.Tax,∃worksFor.Company v Employee
A =
ibm : Company,mary : Parent,
john : EmpStud,(john, ibm) : worksFor
• Is the interpretation I below a model of KB ?
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyIEm
pStu
dI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 24
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise• Let KB = 〈T ,A〉, where:
T =
EmpStud ≡ Student u Employee,Student u ¬Employee v ¬∃pays.Tax,
EmpStud u ¬Parent v ∃pays.Tax,EmpStud u Parent v ¬∃pays.Tax,∃worksFor.Company v Employee
A =
ibm : Company,mary : Parent,
john : EmpStud,(john, ibm) : worksFor
• Is the interpretation I below a model of KB ? Yep!
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyIEm
pStu
dI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 24
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise• Let KB = 〈T ,A〉, where:
T =
EmpStud ≡ Student u Employee,Student u ¬Employee v ¬∃pays.Tax,
EmpStud u ¬Parent v ∃pays.Tax,EmpStud u Parent v ¬∃pays.Tax,∃worksFor.Company v Employee
A =
ibm : Company,mary : Parent,
john : EmpStud,(john, ibm) : worksFor
• Find a counter-model for this knowledge base
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyI
EmpS
tudI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 25
-
Making Statements DL Knowledge Bases Entailment in DLs
Exercise• Let KB = 〈T ,A〉, where:
T =
EmpStud ≡ Student u Employee,Student u ¬Employee v ¬∃pays.Tax,
EmpStud u ¬Parent v ∃pays.Tax,EmpStud u Parent v ¬∃pays.Tax,∃worksFor.Company v Employee
A =
ibm : Company,mary : Parent,
john : EmpStud,(john, ibm) : worksFor
• Find a counter-model for this knowledge base
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyIEm
pStu
dI
x0 x1 x2(mary) x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 25
-
Making Statements DL Knowledge Bases Entailment in DLs
Translating KBs. . .. . . in FOLExtend τ to a mapping from KB = 〈T ,A〉 to conjunctions of FOL formulae
τ(C v D) = ∀x.[τ(C, x)→ τ(D,x)]τ(C ≡ D) = ∀x.[τ(C, x)↔ τ(D,x)]
τ(a : C) = τ(C, ca)τ((a, b) : r) = pr(ca, cb)
Then• τ(KB) = (
∧α∈T τ(α)) ∧ (
∧α∈A τ(α))
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 26
-
Making Statements DL Knowledge Bases Entailment in DLs
Translating KBs. . .. . . in FOLExtend τ to a mapping from KB = 〈T ,A〉 to conjunctions of FOL formulae
τ(C v D) = ∀x.[τ(C, x)→ τ(D,x)]τ(C ≡ D) = ∀x.[τ(C, x)↔ τ(D,x)]
τ(a : C) = τ(C, ca)τ((a, b) : r) = pr(ca, cb)
Then• τ(KB) = (
∧α∈T τ(α)) ∧ (
∧α∈A τ(α))
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 26
-
Making Statements DL Knowledge Bases Entailment in DLs
Translating KBs. . .. . . in FOLExtend τ to a mapping from KB = 〈T ,A〉 to conjunctions of FOL formulae
τ(C v D) = ∀x.[τ(C, x)→ τ(D,x)]τ(C ≡ D) = ∀x.[τ(C, x)↔ τ(D,x)]
τ(a : C) = τ(C, ca)τ((a, b) : r) = pr(ca, cb)
Then• τ(KB) = (
∧α∈T τ(α)) ∧ (
∧α∈A τ(α)) ∧ (
∧distinct a,b occurring in A ca 6= cb)
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 26
-
Making Statements DL Knowledge Bases Entailment in DLs
Translating KBs. . .. . . in FOLExtend τ to a mapping from KB = 〈T ,A〉 to conjunctions of FOL formulae
τ(C v D) = ∀x.[τ(C, x)→ τ(D,x)]τ(C ≡ D) = ∀x.[τ(C, x)↔ τ(D,x)]
τ(a : C) = τ(C, ca)τ((a, b) : r) = pr(ca, cb)
Then• τ(KB) = (
∧α∈T τ(α)) ∧ (
∧α∈A τ(α)) ∧ (
∧distinct a,b occurring in A ca 6= cb)
TheoremKB is satisfiable iff τ(KB) is FOL-satisfiable
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 26
-
Making Statements DL Knowledge Bases Entailment in DLs
Translating KBs. . .. . . in modal logicWe need a universal modality and nominals
ThenWe extend η to a mapping from statements to formulae of hybrid logic
η(C v D) = 2(η(C)→ η(D))η(C ≡ D) = 2(η(C)↔ η(D))
η(a : C) = @naη(C)η((a, b) : r) = @na〈ir〉nb
TheoremKB is satisfiable iff η(KB) is modally satisfiable
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 27
-
Making Statements DL Knowledge Bases Entailment in DLs
Translating KBs. . .. . . in modal logicWe need a universal modality and nominals
ThenWe extend η to a mapping from statements to formulae of hybrid logic
η(C v D) = 2(η(C)→ η(D))η(C ≡ D) = 2(η(C)↔ η(D))
η(a : C) = @naη(C)η((a, b) : r) = @na〈ir〉nb
TheoremKB is satisfiable iff η(KB) is modally satisfiable
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 27
-
Making Statements DL Knowledge Bases Entailment in DLs
Outline
Making Statements
DL Knowledge Bases
Entailment in DLs
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 28
-
Making Statements DL Knowledge Bases Entailment in DLs
What does follow from a KB?
Example• Is an employed parent an employee who is also a student?
• Is being an employee the same as being employed by someone?
• Is Mary an employed parent?
• Is being an employee more general than being a employed student?
• Is there anybody who is a student and a parent at the same time?
• Is my knowledge base consistent?
• Tell me, briefly, what John is.
• Who are the employed students who work for companies?
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 29
-
Making Statements DL Knowledge Bases Entailment in DLs
What does follow from a KB?Entailment from KBs• Defined on the level of statements (not concepts)• Remember:
In many logics
meta-language(entailment, etc)
object language(formulae)
In DLs
meta-language
statements
concept language
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 30
-
Making Statements DL Knowledge Bases Entailment in DLs
What does follow from a KB?Entailment from KBs• Given a TBox T , what other subsumptions follow?• Given an ABox A, what other assertions follow?• Given a knowledge base KB = 〈T ,A〉, what statements follow from it?
Obvious definition of entailment• T |= α if I α for every I s.t. I T• A |= α if I α for every I s.t. I A• KB |= α if I α for every I s.t. I KB
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 31
-
Making Statements DL Knowledge Bases Entailment in DLs
What does follow from a KB?Entailment from KBs• Given a TBox T , what other subsumptions follow?• Given an ABox A, what other assertions follow?• Given a knowledge base KB = 〈T ,A〉, what statements follow from it?
Obvious definition of entailment• T |= α if I α for every I s.t. I T• A |= α if I α for every I s.t. I A• KB |= α if I α for every I s.t. I KB
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 31
-
Making Statements DL Knowledge Bases Entailment in DLs
What does follow from a KB?
Example
T =
EmpStud ≡ Student u Employee,Student u ¬Employee v ¬∃pays.Tax,
EmpStud u ¬Parent v ∃pays.Tax,EmpStud u Parent v ¬∃pays.Tax,∃worksFor.Company v Employee
A =
ibm : Company,mary : Parent,
john : EmpStud,(john, ibm) : worksFor
• KB |= Student u ∃worksFor.Company u ¬Parent v EmpStud u ∃pays.Tax
• KB |= john : Student u ∃worksFor.Company
• KB 6|= mary : ¬∃pays.Tax
• KB 6|= Employee v ∃empBy.Company
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 32
-
Making Statements DL Knowledge Bases Entailment in DLs
What does follow from a KB?
Example• KB 6|= mary : ¬∃pays.Tax• KB 6|= Employee v ∃empBy.Company
I : ∆I
TaxI
ParentI
StudentI EmployeeI
CompanyIEm
pStu
dI
x0 x1(mary) x2 x3
x4 x5(john) x6(ibm)
x7 x8 x9 x10
pays
pays worksFor
worksForempBy
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 33
-
Making Statements DL Knowledge Bases Entailment in DLs
Open- v. closed-world assumptionClosed-world assumption (CWA)• KB contains all information• Non-derivable statements are assumed to be false
Open-world assumption (OWA)• KB may be incomplete• Truth of non-derivable statements is just unknown
Example{(john, ibm) : worksFor, ibm : Company)} |= john : ∀worksFor.Company ?
• In Prolog: “Yep!”• In DL-based systems: “Uh, I don’t know . . . ”
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 34
-
Making Statements DL Knowledge Bases Entailment in DLs
Open- v. closed-world assumptionClosed-world assumption (CWA)• KB contains all information• Non-derivable statements are assumed to be false
Open-world assumption (OWA)• KB may be incomplete• Truth of non-derivable statements is just unknown
Example{(john, ibm) : worksFor, ibm : Company)} |= john : ∀worksFor.Company ?
• In Prolog: “Yep!”• In DL-based systems: “Uh, I don’t know . . . ”
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 34
-
Making Statements DL Knowledge Bases Entailment in DLs
Open- v. closed-world assumptionClosed-world assumption (CWA)• KB contains all information• Non-derivable statements are assumed to be false
Open-world assumption (OWA)• KB may be incomplete• Truth of non-derivable statements is just unknown
Example{(john, ibm) : worksFor, ibm : Company)} |= john : ∀worksFor.Company ?
• In Prolog: “Yep!”• In DL-based systems: “Uh, I don’t know . . . ”
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 34
-
Making Statements DL Knowledge Bases Entailment in DLs
DLs are (very) classical
Definition (Consequence operator)Given KB = 〈T ,A〉, the set of all consequences of KB is
Cn(KB) =def {α | KB |= α}
TheoremCn(·) satisfies the following properties:
• KB ⊆ Cn(KB) (Inclusion)
• Cn(KB) = Cn(Cn(KB)) (Idempotency)
• If KB1 ⊆ KB2, then Cn(KB1) ⊆ Cn(KB2) (Monotonicity)
Hence, Cn(·) is a Tarskian consequence operator
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 35
-
Making Statements DL Knowledge Bases Entailment in DLs
DLs are (very) classical
Definition (Consequence operator)Given KB = 〈T ,A〉, the set of all consequences of KB is
Cn(KB) =def {α | KB |= α}
TheoremCn(·) satisfies the following properties:
• KB ⊆ Cn(KB) (Inclusion)
• Cn(KB) = Cn(Cn(KB)) (Idempotency)
• If KB1 ⊆ KB2, then Cn(KB1) ⊆ Cn(KB2) (Monotonicity)
Hence, Cn(·) is a Tarskian consequence operator
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 35
-
Making Statements DL Knowledge Bases Entailment in DLs
DLs are (very) classical
Definition (Consequence operator)Given KB = 〈T ,A〉, the set of all consequences of KB is
Cn(KB) =def {α | KB |= α}
TheoremCn(·) satisfies the following properties:
• KB ⊆ Cn(KB) (Inclusion)
• Cn(KB) = Cn(Cn(KB)) (Idempotency)
• If KB1 ⊆ KB2, then Cn(KB1) ⊆ Cn(KB2) (Monotonicity)
Hence, Cn(·) is a Tarskian consequence operator
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 35
-
Making Statements DL Knowledge Bases Entailment in DLs
EpilogueSummary• Intensional and extensional knowledge
• Specifying DL knowledge bases• TBox: categories• ABox: partial view of the world
• What follows from a DL ontology
What next?• Reasoning with DL ontologies
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 36
-
Making Statements DL Knowledge Bases Entailment in DLs
EpilogueSummary• Intensional and extensional knowledge
• Specifying DL knowledge bases• TBox: categories• ABox: partial view of the world
• What follows from a DL ontology
What next?• Reasoning with DL ontologies
Ivan Varzinczak (CRIL) An Introduction to Description Logics (Part 2) ESSLLI 2018 36
Making StatementsDL Knowledge BasesEntailment in DLs