SlideShare une entreprise Scribd logo
1  sur  55
Télécharger pour lire hors ligne
Extending OWL with Integrity Constraints

                Jiao Tao1 , Evren Sirin2 , Jie Bao1 , Deborah L. McGuinness1
                      taoj2@cs.rpi.edu, evren@clarkparsia.com
                         baojie@cs.rpi.edu, dlm@cs.rpi.edu
                                           1 Tetherless World Constellation,

                                          Department of Computer Science,
                                           Rensselaer Polytechnic Institute,
                                                Troy, NY 12180, USA.
                                                   2 Clark & Parsia,

                                            Washington, DC 20001, USA.


          The 23rd International Workshop on Description Logics (DL 2010)



Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   1 / 20
Outline



           Background
                   OWA and nUNA in OWL
                   IC Requirements for OWL
                   Related Work
                   Contributions
           Preliminaries
                   SROIQ
                   DCQ
           IC Semantics for OWL
                   Epistemic Approach
                   Formalization of IC Semantics
           IC Validation
                   DCQnot
                   From ICs to DCQnot
                   IC Validation Algorithm
           Implementation
           Conclusions and Future Work
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   2 / 20
Background     OWA and nUNA in OWL


   OWA and nUNA in OWL


           Open World Assumption (OWA)
                   CWA vs. OWA: any statement that is not known to be true is false vs.
                   unknown
                   Example 1: K satisfies α although p does not have a known producer.
                   K = {Product(p)}, α : Product ∃hasProducer.Producer




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints    May 2010, DL 2010   3 / 20
Background     OWA and nUNA in OWL


   OWA and nUNA in OWL


           Open World Assumption (OWA)
                   CWA vs. OWA: any statement that is not known to be true is false vs.
                   unknown
                   Example 1: K satisfies α although p does not have a known producer.
                   K = {Product(p)}, α : Product ∃hasProducer.Producer
           nonUnique Name Assumption (nUNA)
                   UNA vs. nUNA: two different names always refer to different entities
                   vs. may refer to the same entity
                   Example 2: K satisfies α although p has two producers.
                   K = {Product(p), hasProducer(p, m1 ), hasProducer(p, m2 )}
                   α : Product ≤ 1hasProducer. .




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints    May 2010, DL 2010   3 / 20
Background     OWA and nUNA in OWL


   OWA and nUNA in OWL


           Open World Assumption (OWA)
                   CWA vs. OWA: any statement that is not known to be true is false vs.
                   unknown
                   Example 1: K satisfies α although p does not have a known producer.
                   K = {Product(p)}, α : Product ∃hasProducer.Producer
           nonUnique Name Assumption (nUNA)
                   UNA vs. nUNA: two different names always refer to different entities
                   vs. may refer to the same entity
                   Example 2: K satisfies α although p has two producers.
                   K = {Product(p), hasProducer(p, m1 ), hasProducer(p, m2 )}
                   α : Product ≤ 1hasProducer. .
           Difficult to use OWL for integrity constraint(IC) validation!


Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints    May 2010, DL 2010   3 / 20
Background     IC Requirements for OWL


   IC Requirements for OWL


           Strong needs to use OWL as an IC language! Especially in hybrid
           settings where OWL KBs are integrated with relational databases and
           ICs are needed to enforce the named individuals in KBs to have some
           known values.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   4 / 20
Background     IC Requirements for OWL


   IC Requirements for OWL


           Strong needs to use OWL as an IC language! Especially in hybrid
           settings where OWL KBs are integrated with relational databases and
           ICs are needed to enforce the named individuals in KBs to have some
           known values.
                   Typing Constraints
                           Individuals participating in a relation should be instances of certain
                           types, i.e., domain&range axioms and universal restrictions.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   4 / 20
Background     IC Requirements for OWL


   IC Requirements for OWL


           Strong needs to use OWL as an IC language! Especially in hybrid
           settings where OWL KBs are integrated with relational databases and
           ICs are needed to enforce the named individuals in KBs to have some
           known values.
                   Typing Constraints
                           Individuals participating in a relation should be instances of certain
                           types, i.e., domain&range axioms and universal restrictions.
                   Participation Constraints
                           Instances of the constrained class should have a role assertion i.e.,
                           existential restrictions.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   4 / 20
Background     IC Requirements for OWL


   IC Requirements for OWL


           Strong needs to use OWL as an IC language! Especially in hybrid
           settings where OWL KBs are integrated with relational databases and
           ICs are needed to enforce the named individuals in KBs to have some
           known values.
                   Typing Constraints
                           Individuals participating in a relation should be instances of certain
                           types, i.e., domain&range axioms and universal restrictions.
                   Participation Constraints
                           Instances of the constrained class should have a role assertion i.e.,
                           existential restrictions.
                   Uniqueness Constraints
                           An individual cannot participate in multiple role assertions with the
                           same role, i.e., functional property restrictions, cardinality restrictions.



Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   4 / 20
Background     Related Work


   Related Work


           Rule-based approach [4, 6]
                   ⊥ ← DL[Product](x), not P(x, y )
                   P(x, y ) ← DL[hasProducer](x, y ), DL[Producer](y )
                   IC validation: KDL , ICRule |= ⊥
                   Limitations: ontology developers need to be familiar with rules




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   5 / 20
Background     Related Work


   Related Work


           Rule-based approach [4, 6]
                   ⊥ ← DL[Product](x), not P(x, y )
                   P(x, y ) ← DL[hasProducer](x, y ), DL[Producer](y )
                   IC validation: KDL , ICRule |= ⊥
                   Limitations: ontology developers need to be familiar with rules
           Epistemic query-based approach [1]
                   KProduct(x) → ∃y .(KhasProducer(x, y ) ∧ KProducer(y ))
                   IC validation: Ans(ICEQL−Lite , KFOL )
                   Limitations: complexity results in expressive DLs are unknown




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   5 / 20
Background     Related Work


   Related Work


           Rule-based approach [4, 6]
                   ⊥ ← DL[Product](x), not P(x, y )
                   P(x, y ) ← DL[hasProducer](x, y ), DL[Producer](y )
                   IC validation: KDL , ICRule |= ⊥
                   Limitations: ontology developers need to be familiar with rules
           Epistemic query-based approach [1]
                   KProduct(x) → ∃y .(KhasProducer(x, y ) ∧ KProducer(y ))
                   IC validation: Ans(ICEQL−Lite , KFOL )
                   Limitations: complexity results in expressive DLs are unknown
           Approach based on epistemic extension of DLs [2, 3]
                   KProduct ∃KhasProducer.KProducer
                   IC validation: KDL |= ICEDL
                   Limitations: less expressive DLs, strict UNA



Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   5 / 20
Background     Related Work cont.


   Related Work cont.


           Approach reuses OWL as an IC language




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   6 / 20
Background     Related Work cont.


   Related Work cont.


           Approach reuses OWL as an IC language
                   IC validation: T , A |=MM C




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   6 / 20
Background     Related Work cont.


   Related Work cont.


           Approach reuses OWL as an IC language
                   IC validation: T , A |=MM C
                   Limitations:




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   6 / 20
Background     Related Work cont.


   Related Work cont.


           Approach reuses OWL as an IC language
                   IC validation: T , A |=MM C
                   Limitations:
                           Unintuitive
                           Example 4: α is satisfied although the producer of p is unknown.
                           K = {Product(p), ∃hasProducer.Producer(p)}
                           α : Product ∃hasProducer.Producer




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   6 / 20
Background     Related Work cont.


   Related Work cont.


           Approach reuses OWL as an IC language
                   IC validation: T , A |=MM C
                   Limitations:
                           Unintuitive
                           Example 4: α is satisfied although the producer of p is unknown.
                           K = {Product(p), ∃hasProducer.Producer(p)}
                           α : Product ∃hasProducer.Producer
                           Maintenance burden
                           Example 5: α is satisfied although we are not sure if the producer of p
                           is m1 or m2 .
                           K = {Product(p), (∃hasProducer.{m1 , m2 })(p), Producer(m1 ),
                                Producer(m2 ), O(p), O(m1 ), O(m2 )}
                           α : Product ∃hasProducer.(Producer O)




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   6 / 20
Background     Contributions


   Contributions




   The main contributions of this paper include:
           An IC semantics for OWL
           A sound and complete IC validation algorithm
           A prototype implementation




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   7 / 20
Preliminaries   SROIQ


   Description Logics SROIQ [5]

           SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset
           of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an
           element of ∆.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   8 / 20
Preliminaries   SROIQ


   Description Logics SROIQ [5]

           SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset
           of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an
           element of ∆.
           I satisfies an axiom α, denoted I |= α, if
                               Type             Axiom                   Condition on I
                               TBox             C D                         C I ⊆ DI
                                              R1 R2                            I
                                                                            R1 ⊆ R2     I

                                            R1 . . . Rn R               I ◦ . . . ◦ RI ⊆ RI
                                                                       R1             n
                               RBox            Ref(R)                 ∀x ∈ ∆ : x, x ∈ R I
                                               Irr(R)                 ∀x ∈ ∆ : x, x ∈ R I
                                             Dis(R1 , R2 )                 I        I
                                                                          R1 ∩ R2 = ∅




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   8 / 20
Preliminaries   SROIQ


   Description Logics SROIQ [5]

           SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset
           of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an
           element of ∆.
           I satisfies an axiom α, denoted I |= α, if
                               Type             Axiom                   Condition on I
                               TBox             C D                         C I ⊆ DI
                                              R1 R2                            I
                                                                            R1 ⊆ R2     I

                                            R1 . . . Rn R               I ◦ . . . ◦ RI ⊆ RI
                                                                       R1             n
                               RBox            Ref(R)                 ∀x ∈ ∆ : x, x ∈ R I
                                               Irr(R)                 ∀x ∈ ∆ : x, x ∈ R I
                                             Dis(R1 , R2 )                 I        I
                                                                          R1 ∩ R2 = ∅

           I is a model of K if it satisfies all the axioms in K.


Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   8 / 20
Preliminaries   SROIQ


   Description Logics SROIQ [5]

           SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset
           of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an
           element of ∆.
           I satisfies an axiom α, denoted I |= α, if
                               Type             Axiom                   Condition on I
                               TBox             C D                         C I ⊆ DI
                                              R1 R2                            I
                                                                            R1 ⊆ R2     I

                                            R1 . . . Rn R               I ◦ . . . ◦ RI ⊆ RI
                                                                       R1             n
                               RBox            Ref(R)                 ∀x ∈ ∆ : x, x ∈ R I
                                               Irr(R)                 ∀x ∈ ∆ : x, x ∈ R I
                                             Dis(R1 , R2 )                 I        I
                                                                          R1 ∩ R2 = ∅

           I is a model of K if it satisfies all the axioms in K.
           K entails α, written as K |= α, if I |= α for all models I ∈ Mod(K).

Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   8 / 20
Preliminaries   DCQ


   Distinguished Conjunctive Query (DCQ)

           Syntax:
                   Query atom: q ← C (x) | R(x, y ) | ¬R(x, y ) | x = y | x = y
                   Conjunctive Query (CQ): Q ← q | Q1 ∧ Q2
                   Distinguished Conjunctive Query (DCQ): CQ containing only
                   distinguished variables1




      1
        A distinguished variable can be mapped to only known individuals, i.e., an element
   from NI .
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   9 / 20
Preliminaries    DCQ


   Distinguished Conjunctive Query (DCQ)

           Syntax:
                   Query atom: q ← C (x) | R(x, y ) | ¬R(x, y ) | x = y | x = y
                   Conjunctive Query (CQ): Q ← q | Q1 ∧ Q2
                   Distinguished Conjunctive Query (DCQ): CQ containing only
                   distinguished variables1
           Semantics:
                   Given a KB K, a DCQ Q, and an assignment σ : NV → NI , we say
                   that K |=σ Q if:

                              K |=σ q                             iff         K |= σ(q)
                                     σ
                              K |= Q1 ∧ Q2                        iff         K |=σ Q1 and K |=σ Q2

                   where σ(Q) denotes the application of σ to Q.
                   Ans(Q, K) = {σ | K |=σ Q}
                   K |= Q if Ans(Q, K) = ∅
      1
        A distinguished variable can be mapped to only known individuals, i.e., an element
   from NI .
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints     May 2010, DL 2010   9 / 20
IC Semantics for OWL     Epistemic Approach


   Epistemic Approach




           ICs are epistemic in nature [8]: about what the KB knows.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   10 / 20
IC Semantics for OWL     Epistemic Approach


   Epistemic Approach




           ICs are epistemic in nature [8]: about what the KB knows.
           We adopt this epistemic view of ICs and propose an IC semantics for
           OWL, which is similar to how the semantics of epistemic DL ALCK
           [2] and MKNF DL ALCKN F [3] are defined, but is still different from
           them:




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   10 / 20
IC Semantics for OWL     Epistemic Approach


   Epistemic Approach




           ICs are epistemic in nature [8]: about what the KB knows.
           We adopt this epistemic view of ICs and propose an IC semantics for
           OWL, which is similar to how the semantics of epistemic DL ALCK
           [2] and MKNF DL ALCKN F [3] are defined, but is still different from
           them:
                   weak UNA vs. strict UNA
                   expressive DLs, i.e., SROIQ, vs. less expressive DLs, i.e., ALC.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   10 / 20
IC Semantics for OWL     Formalization of IC Semantics


   IC-interpretation

           An IC-interpretation is a pair I, U where I is a SROIQ
           interpretation defined over ∆I and U is a set of SROIQ
           interpretations.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   11 / 20
IC Semantics for OWL     Formalization of IC Semantics


   IC-interpretation

           An IC-interpretation is a pair I, U where I is a SROIQ
           interpretation defined over ∆I and U is a set of SROIQ
           interpretations.
           IC-interpretations of concepts, roles, individuals:

           C I,U = {x I | x ∈ NI s.t. ∀J ∈ U, x J ∈ C J }
           R I,U = { x I , y I | x, y ∈ NI s.t. ∀J ∈ U, x J , y J ∈ R J }
           (R − )I,U = { x I , y I | y I , x I ∈ R I,U }
           (C      D)I,U = C I,U ∩ D I,U ,               (¬C )I,U = NI  C I,U
           (≥ nR.C )I,U = {x I | x ∈ NI s.t. #{y I | x I , y I ∈ R I,U , y I ∈ C I,U } ≥ n}
           (∃R.Self)I,U = {x I | x ∈ NI s.t. x I , x I ∈ R I,U },                          {a}I,U = {aI }.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   11 / 20
IC Semantics for OWL     Formalization of IC Semantics


   IC-interpretation

           An IC-interpretation is a pair I, U where I is a SROIQ
           interpretation defined over ∆I and U is a set of SROIQ
           interpretations.
           IC-interpretations of concepts, roles, individuals:

           C I,U = {x I | x ∈ NI s.t. ∀J ∈ U, x J ∈ C J }
           R I,U = { x I , y I | x, y ∈ NI s.t. ∀J ∈ U, x J , y J ∈ R J }
           (R − )I,U = { x I , y I | y I , x I ∈ R I,U }
           (C      D)I,U = C I,U ∩ D I,U ,               (¬C )I,U = NI  C I,U
           (≥ nR.C )I,U = {x I | x ∈ NI s.t. #{y I | x I , y I ∈ R I,U , y I ∈ C I,U } ≥ n}
           (∃R.Self)I,U = {x I | x ∈ NI s.t. x I , x I ∈ R I,U },                          {a}I,U = {aI }.

           IC-interpretations use the CWA w.r.t. NI (standard DL
           interpretations: OWA).

Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   11 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Weak UNA


           Weak UNA: two named individuals with different identifiers are
           assumed to be different by default unless their equality is required to
           satisfy the axioms in the KB.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   12 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Weak UNA


           Weak UNA: two named individuals with different identifiers are
           assumed to be different by default unless their equality is required to
           satisfy the axioms in the KB.
           How weak UNA is formalized?




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   12 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Weak UNA


           Weak UNA: two named individuals with different identifiers are
           assumed to be different by default unless their equality is required to
           satisfy the axioms in the KB.
           How weak UNA is formalized?
                   J      =   I if:
                           For every concept C, J |= C (a) implies I |= C (a);
                           For every role R, J |= R(a, b) implies I |= R(a, b);
                           EJ ⊂ EI
                   where EI = { a, b | a, b ∈ NI s.t. I |= a = b}.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   12 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Weak UNA


           Weak UNA: two named individuals with different identifiers are
           assumed to be different by default unless their equality is required to
           satisfy the axioms in the KB.
           How weak UNA is formalized?
                   J      =   I if:
                           For every concept C, J |= C (a) implies I |= C (a);
                           For every role R, J |= R(a, b) implies I |= R(a, b);
                           EJ ⊂ EI
                   where EI = { a, b | a, b ∈ NI s.t. I |= a = b}.
                   ME models: the models of K with minimal equality (ME) between NI :
                   ModME (K) = {I ∈ Mod(K) | J , J ∈ Mod(K), J = I}




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   12 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Weak UNA


           Weak UNA: two named individuals with different identifiers are
           assumed to be different by default unless their equality is required to
           satisfy the axioms in the KB.
           How weak UNA is formalized?
                   J      =   I if:
                           For every concept C, J |= C (a) implies I |= C (a);
                           For every role R, J |= R(a, b) implies I |= R(a, b);
                           EJ ⊂ EI
                   where EI = { a, b | a, b ∈ NI s.t. I |= a = b}.
                   ME models: the models of K with minimal equality (ME) between NI :
                   ModME (K) = {I ∈ Mod(K) | J , J ∈ Mod(K), J = I}
           IC-semantics uses the weak UNA (standard DL semantics: nUNA).



Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   12 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Axiom Satisfaction, IC Satisfaction, Extended KB
           An IC-interpretation I, U satisfies an axiom α, denoted as I, U |= α,if
                        Type             Axiom                      Condition on I, U
                        TBox             C D                           C I,U ⊆ D I,U
                                                                         I,U        I,U
                                       R1 R2                            R1 ⊆ R2
                                                                   I,U           I,U
                                     R1 . . . Rn R                R1 ◦ . . . ◦ Rn ⊆ R I,U
                        RBox            Ref(R)                  ∀x ∈ NI : x I,U , x I,U ∈ R I,U
                                        Irr(R)                  ∀x ∈ NI : x I,U , x I,U ∈ R I,U
                                                                       I,U      I,U
                                      Dis(R1 , R2 )                   R1 ∩ R2 = ∅




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   13 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Axiom Satisfaction, IC Satisfaction, Extended KB
           An IC-interpretation I, U satisfies an axiom α, denoted as I, U |= α,if
                        Type             Axiom                      Condition on I, U
                        TBox             C D                           C I,U ⊆ D I,U
                                                                         I,U        I,U
                                       R1 R2                            R1 ⊆ R2
                                                                   I,U           I,U
                                     R1 . . . Rn R                R1 ◦ . . . ◦ Rn ⊆ R I,U
                        RBox            Ref(R)                  ∀x ∈ NI : x I,U , x I,U ∈ R I,U
                                        Irr(R)                  ∀x ∈ NI : x I,U , x I,U ∈ R I,U
                                                                       I,U      I,U
                                      Dis(R1 , R2 )                   R1 ∩ R2 = ∅
           The IC-satisfaction of α by K, i.e., K |=IC α, is defined as:
                        K |=IC α iff ∀I ∈ U, I, U |= α, where U = ModME (K)




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   13 / 20
IC Semantics for OWL     Formalization of IC Semantics


   Axiom Satisfaction, IC Satisfaction, Extended KB
           An IC-interpretation I, U satisfies an axiom α, denoted as I, U |= α,if
                        Type             Axiom                      Condition on I, U
                        TBox             C D                           C I,U ⊆ D I,U
                                                                         I,U        I,U
                                       R1 R2                            R1 ⊆ R2
                                                                   I,U           I,U
                                     R1 . . . Rn R                R1 ◦ . . . ◦ Rn ⊆ R I,U
                        RBox            Ref(R)                  ∀x ∈ NI : x I,U , x I,U ∈ R I,U
                                        Irr(R)                  ∀x ∈ NI : x I,U , x I,U ∈ R I,U
                                                                       I,U      I,U
                                      Dis(R1 , R2 )                   R1 ∩ R2 = ∅
           The IC-satisfaction of α by K, i.e., K |=IC α, is defined as:
                        K |=IC α iff ∀I ∈ U, I, U |= α, where U = ModME (K)
           An extended KB is a pair K, C where K is a SROIQ KB
           interpreted with the standard OWL semantics and C is a set of
           SROIQ axioms interpreted with IC semantics. We say that K, C is
           valid if ∀α ∈ C, K |=IC α.
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints              May 2010, DL 2010   13 / 20
IC Validation   DCQnot


   Distinguished Conjunctive Query with not (DCQnot )



           DCQnot is the extension of DCQ with the NAF operator not .




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   14 / 20
IC Validation    DCQnot


   Distinguished Conjunctive Query with not (DCQnot )



           DCQnot is the extension of DCQ with the NAF operator not .
           Syntax: Q ← q | Q1 ∧ Q2 | not Q | Q1 ∨ Q2
           where Q1 ∨ Q2 is the abbreviation for not (not Q1 ∧ not Q2 ).
           Semantics:

                       K |=σ not Q                           iff                  σ s.t. K |=σ σ(Q)
                       K |=σ Q1 ∨ Q2                          iff            K |=σ Q1 or K |=σ Q2




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints    May 2010, DL 2010   14 / 20
IC Validation   From ICs to DCQnot


   From ICs to DCQnot


           Translation from ICs to DCQnot : translate an IC axiom α into a
           query Q such that when α is satisfied, Q is false.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   15 / 20
IC Validation   From ICs to DCQnot


   From ICs to DCQnot


           Translation from ICs to DCQnot : translate an IC axiom α into a
           query Q such that when α is satisfied, Q is false.
           Translation rules:
                   Concept Translation:

                       Tc (Ca , x) := Ca (x)
                       Tc (¬C , x) := not Tc (C , x)
                       Tc (C1      C2 , x) := Tc (C1 , x) ∧ Tc (C2 , x)
                       Tc (≥ nR.C , x) :=                 (R(x, yi ) ∧ Tc (C , yi ))             not (yi = yj )
                                                  1≤i≤n                                1≤i<j≤n

                       Tc (∃R.Self, x) := R(x, x)
                       Tc ({a}, x) := (x = a)



Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints      May 2010, DL 2010   15 / 20
IC Validation   From ICs to DCQnot cont.


   From ICs to DCQnot cont.
           Translation rules cont.:
                   Axiom Translation:
                       T (C1 C2 ) := Tc (C1 , x) ∧ not Tc (C2 , x)
                         T (R1        R2 ) := R1 (x, y ) ∧ not R2 (x, y )
                         T (R1 . . . Rn        R) := R1 (x, y1 ) ∧ . . . Rn (yn−1 , yn ) ∧ not R(x, yn )
                         T (Ref(R)) := not R(x, x)
                         T (Irr(R)) := R(x, x)
                         T (Dis(R1 , R2 )) := R1 (x, y ) ∧ R2 (x, y )




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints         May 2010, DL 2010   16 / 20
IC Validation   From ICs to DCQnot cont.


   From ICs to DCQnot cont.
           Translation rules cont.:
                   Axiom Translation:
                       T (C1 C2 ) := Tc (C1 , x) ∧ not Tc (C2 , x)
                         T (R1        R2 ) := R1 (x, y ) ∧ not R2 (x, y )
                         T (R1 . . . Rn        R) := R1 (x, y1 ) ∧ . . . Rn (yn−1 , yn ) ∧ not R(x, yn )
                         T (Ref(R)) := not R(x, x)
                         T (Irr(R)) := R(x, x)
                         T (Dis(R1 , R2 )) := R1 (x, y ) ∧ R2 (x, y )
           Example 6:
                 T (Product              ∃hasProducer.Producer)
                  := Tc (Product, x) ∧ not Tc (∃hasProducer.Producer, x)
                  := Product(x) ∧ not (hasProducer(x, y ) ∧ Tc (Producer, y ))
                  := Product(x) ∧ not (hasProducer(x, y ) ∧ Producer(y ))

Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints         May 2010, DL 2010   16 / 20
IC Validation   IC Validation Algorithm


   IC Validation Algorithm



   Theorem: Given an extended KB K, C with expressivity SRI, SROIQ
   ( SROIQ, SROI resp.), we have that K |=IC α iff K |= T (α) where
   α ∈ C [9].




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   17 / 20
IC Validation   IC Validation Algorithm


   IC Validation Algorithm



   Theorem: Given an extended KB K, C with expressivity SRI, SROIQ
   ( SROIQ, SROI resp.), we have that K |=IC α iff K |= T (α) where
   α ∈ C [9].


   To limit the interactions between the disjunctive (in)equality axioms in K
   and the cardinality restrictions in C, we can require:




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   17 / 20
IC Validation   IC Validation Algorithm


   IC Validation Algorithm



   Theorem: Given an extended KB K, C with expressivity SRI, SROIQ
   ( SROIQ, SROI resp.), we have that K |=IC α iff K |= T (α) where
   α ∈ C [9].


   To limit the interactions between the disjunctive (in)equality axioms in K
   and the cardinality restrictions in C, we can require:
        K: within SRI, no nominals & cardinality restrictions thus no
        disjunctive (in)equality;
           C: within SROI, no cardinality restrictions.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints        May 2010, DL 2010   17 / 20
Implementation


   Implementation


             Prototype: IC validator              2   in the OWL 2 DL reasoner Pellet.
             SPARQL-based [7] implementation.




        2
            http://clarkparsia.com/pellet/oicv-0.1.2.zip
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   18 / 20
Implementation


   Implementation


             Prototype: IC validator              2   in the OWL 2 DL reasoner Pellet.
             SPARQL-based [7] implementation.
             The validation process is as follows:
                   Reading OWL IC axioms
                   Translating to DCQnot
                   Translating to SPARQL query
                   Executed by the SPARQL query engine




        2
            http://clarkparsia.com/pellet/oicv-0.1.2.zip
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   18 / 20
Implementation


   Implementation


             Prototype: IC validator              2   in the OWL 2 DL reasoner Pellet.
             SPARQL-based [7] implementation.
             The validation process is as follows:
                   Reading OWL IC axioms
                   Translating to DCQnot
                   Translating to SPARQL query
                   Executed by the SPARQL query engine
             Preliminary evaluation:
                   Dataset: LUBM(5), 100K individuals, 800K ABox axioms
                   IC validation time: 2s
                   Logical consistency checking time: 10s



        2
            http://clarkparsia.com/pellet/oicv-0.1.2.zip
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   18 / 20
Conclusions and Future Work


   Conclusions and Future Work




           In this paper, we have
                   described an IC semantics for OWL
                   presented a sound and complete IC validation algorithm
                   described the prototype IC validator




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   19 / 20
Conclusions and Future Work


   Conclusions and Future Work




           In this paper, we have
                   described an IC semantics for OWL
                   presented a sound and complete IC validation algorithm
                   described the prototype IC validator
           In the future, we will:
                   look at the performance of IC validation in real-world datasets
                   explore the IC validation algorithms for the fullly expressive DLs




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   19 / 20
Questions




                                             Questions?


                   Thanks for your attention!



Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   20 / 20
Questions


          D. Calvanese, G. D. Giacomo, D. Lembo, M. Lenzerini, and R. Rosati.
          EQL-Lite: Effective First-Order Query Processing in Description
          Logics.
          In Proc. IJCAI2007, pages 274–279, 2007.
          F. M. Donini, M. Lenzerini, D. Nardi, W. Nutt, and A. Schaerf.
          An Epistemic Operator for Description Logics.
          AI, 100(1–2):225–274, 1998.
          F. M. Donini, D. Nardi, and R. Rosati.
          Description Logics of Minimal Knowledge and Negation as Failure.
          ACM Trans. Comput. Logic, 3(2):177–225, 2002.
          T. Eiter, G. Ianni, T. Lukasiewicz, R. Schindlauer, and H. Tompits.
          Combining Answer Set Programming with Description Logics for the
          Semantic Web.
          AI, 172(12-13):1495–1539, 2008.
          I. Horrocks, O. Kutz, and U. Sattler.
          The Even More Irresistible SROIQ.
Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   20 / 20
Questions

          In Proc. KR2006, pages 57–67, 2006.
          B. Motik.
          A Faithful Integration of Description Logics with Logic Programming.
          In Proc. IJCAI2007, pages 477–482, 2007.
          E. Prud’hommeaux and A. Seaborne.
          SPARQL Query Language for RDF, 2008.
          R. Reiter.
          On Integrity Constraints.
          In Proc. TARK1988, pages 97–111, 1988.
          J. Tao, E. Sirin, J. Bao, and D. McGuinness.
          Integrity Constraints in OWL.
          Technical report, Dept. of Computer Science, Rensselaer Polytechnic
          Institute, Troy, NY., 2010.




Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints   May 2010, DL 2010   20 / 20

Contenu connexe

En vedette

Startup best practices
Startup best practicesStartup best practices
Startup best practicesJie Bao
 
Bai thuc hanh 1 4
Bai thuc hanh 1 4Bai thuc hanh 1 4
Bai thuc hanh 1 4Gió Lạnh
 
Representing financial reports on the semantic web a faithful translation f...
Representing financial reports on the semantic web   a faithful translation f...Representing financial reports on the semantic web   a faithful translation f...
Representing financial reports on the semantic web a faithful translation f...Jie Bao
 
Package-based Description Logics – Preliminary Results
Package-based Description Logics – Preliminary ResultsPackage-based Description Logics – Preliminary Results
Package-based Description Logics – Preliminary ResultsJie Bao
 
Expressive Query Answering For Semantic Wikis
Expressive Query Answering For  Semantic WikisExpressive Query Answering For  Semantic Wikis
Expressive Query Answering For Semantic WikisJie Bao
 
On the Decidability of Role Mappings Between Modular Ontologies
On the Decidability of Role Mappings Between Modular OntologiesOn the Decidability of Role Mappings Between Modular Ontologies
On the Decidability of Role Mappings Between Modular OntologiesJie Bao
 
Expressive Query Answering For Semantic Wikis (20min)
Expressive Query Answering For  Semantic Wikis (20min)Expressive Query Answering For  Semantic Wikis (20min)
Expressive Query Answering For Semantic Wikis (20min)Jie Bao
 

En vedette (7)

Startup best practices
Startup best practicesStartup best practices
Startup best practices
 
Bai thuc hanh 1 4
Bai thuc hanh 1 4Bai thuc hanh 1 4
Bai thuc hanh 1 4
 
Representing financial reports on the semantic web a faithful translation f...
Representing financial reports on the semantic web   a faithful translation f...Representing financial reports on the semantic web   a faithful translation f...
Representing financial reports on the semantic web a faithful translation f...
 
Package-based Description Logics – Preliminary Results
Package-based Description Logics – Preliminary ResultsPackage-based Description Logics – Preliminary Results
Package-based Description Logics – Preliminary Results
 
Expressive Query Answering For Semantic Wikis
Expressive Query Answering For  Semantic WikisExpressive Query Answering For  Semantic Wikis
Expressive Query Answering For Semantic Wikis
 
On the Decidability of Role Mappings Between Modular Ontologies
On the Decidability of Role Mappings Between Modular OntologiesOn the Decidability of Role Mappings Between Modular Ontologies
On the Decidability of Role Mappings Between Modular Ontologies
 
Expressive Query Answering For Semantic Wikis (20min)
Expressive Query Answering For  Semantic Wikis (20min)Expressive Query Answering For  Semantic Wikis (20min)
Expressive Query Answering For Semantic Wikis (20min)
 

Similaire à Extending OWL with Integrity Constraints

Probabilistic Abductive Logic Programming using Possible Worlds
Probabilistic Abductive Logic Programming using Possible WorldsProbabilistic Abductive Logic Programming using Possible Worlds
Probabilistic Abductive Logic Programming using Possible WorldsFulvio Rotella
 
OwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxes
OwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxesOwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxes
OwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxesRokan Uddin Faruqui
 
OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...
OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...
OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...Adila Krisnadhi
 
A scalable ontology reasoner via incremental materialization
A scalable ontology reasoner via incremental materializationA scalable ontology reasoner via incremental materialization
A scalable ontology reasoner via incremental materializationRokan Uddin Faruqui
 
Finding common ground: integrating the eagle-i and VIVO ontologies
Finding common ground: integrating the eagle-i and VIVO ontologiesFinding common ground: integrating the eagle-i and VIVO ontologies
Finding common ground: integrating the eagle-i and VIVO ontologiesmhaendel
 
Formalization and implementation of BFO 2 with a focus on the OWL implementation
Formalization and implementation of BFO 2 with a focus on the OWL implementationFormalization and implementation of BFO 2 with a focus on the OWL implementation
Formalization and implementation of BFO 2 with a focus on the OWL implementationgolpedegato2
 
Triplifier talk
Triplifier talkTriplifier talk
Triplifier talkJohn Deck
 
Structural Systems Pharmacology
Structural Systems PharmacologyStructural Systems Pharmacology
Structural Systems PharmacologyPhilip Bourne
 

Similaire à Extending OWL with Integrity Constraints (8)

Probabilistic Abductive Logic Programming using Possible Worlds
Probabilistic Abductive Logic Programming using Possible WorldsProbabilistic Abductive Logic Programming using Possible Worlds
Probabilistic Abductive Logic Programming using Possible Worlds
 
OwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxes
OwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxesOwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxes
OwlOntDB: A Scalable Reasoning System for OWL 2 RL Ontologies with Large ABoxes
 
OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...
OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...
OWLED 2015 - "OWL Capabilities and Limitation: Typecasting and Ontology Patte...
 
A scalable ontology reasoner via incremental materialization
A scalable ontology reasoner via incremental materializationA scalable ontology reasoner via incremental materialization
A scalable ontology reasoner via incremental materialization
 
Finding common ground: integrating the eagle-i and VIVO ontologies
Finding common ground: integrating the eagle-i and VIVO ontologiesFinding common ground: integrating the eagle-i and VIVO ontologies
Finding common ground: integrating the eagle-i and VIVO ontologies
 
Formalization and implementation of BFO 2 with a focus on the OWL implementation
Formalization and implementation of BFO 2 with a focus on the OWL implementationFormalization and implementation of BFO 2 with a focus on the OWL implementation
Formalization and implementation of BFO 2 with a focus on the OWL implementation
 
Triplifier talk
Triplifier talkTriplifier talk
Triplifier talk
 
Structural Systems Pharmacology
Structural Systems PharmacologyStructural Systems Pharmacology
Structural Systems Pharmacology
 

Plus de Jie Bao

python-graph-lovestory
python-graph-lovestorypython-graph-lovestory
python-graph-lovestoryJie Bao
 
unix toolbox 中文版
unix toolbox 中文版unix toolbox 中文版
unix toolbox 中文版Jie Bao
 
unixtoolbox.book
unixtoolbox.bookunixtoolbox.book
unixtoolbox.bookJie Bao
 
Lean startup 精益创业 新创企业的成长思维
Lean startup 精益创业 新创企业的成长思维Lean startup 精益创业 新创企业的成长思维
Lean startup 精益创业 新创企业的成长思维Jie Bao
 
Towards social webtops using semantic wiki
Towards social webtops using semantic wikiTowards social webtops using semantic wiki
Towards social webtops using semantic wikiJie Bao
 
Semantic information theory in 20 minutes
Semantic information theory in 20 minutesSemantic information theory in 20 minutes
Semantic information theory in 20 minutesJie Bao
 
Towards a theory of semantic communication
Towards a theory of semantic communicationTowards a theory of semantic communication
Towards a theory of semantic communicationJie Bao
 
Owl 2 quick reference card a4 size
Owl 2 quick reference card a4 sizeOwl 2 quick reference card a4 size
Owl 2 quick reference card a4 sizeJie Bao
 
ISWC 2010 Metadata Work Summary
ISWC 2010 Metadata Work SummaryISWC 2010 Metadata Work Summary
ISWC 2010 Metadata Work SummaryJie Bao
 
24 Ways to Explore ISWC 2010 Data
24 Ways to Explore ISWC 2010 Data24 Ways to Explore ISWC 2010 Data
24 Ways to Explore ISWC 2010 DataJie Bao
 
Semantic Web: In Quest for the Next Generation Killer Apps
Semantic Web: In Quest for the Next Generation Killer AppsSemantic Web: In Quest for the Next Generation Killer Apps
Semantic Web: In Quest for the Next Generation Killer AppsJie Bao
 
XACML 3.0 (Partial) Concept Map
XACML 3.0 (Partial) Concept MapXACML 3.0 (Partial) Concept Map
XACML 3.0 (Partial) Concept MapJie Bao
 
Development of a Controlled Natural Language Interface for Semantic MediaWiki
Development of a Controlled Natural Language Interface for Semantic MediaWikiDevelopment of a Controlled Natural Language Interface for Semantic MediaWiki
Development of a Controlled Natural Language Interface for Semantic MediaWikiJie Bao
 
Digital image self-adaptive acquisition in medical x-ray imaging
Digital image self-adaptive acquisition in medical x-ray imagingDigital image self-adaptive acquisition in medical x-ray imaging
Digital image self-adaptive acquisition in medical x-ray imagingJie Bao
 
Privacy-Preserving Reasoning on the Semantic Web (Poster)
Privacy-Preserving Reasoning on the Semantic Web (Poster)Privacy-Preserving Reasoning on the Semantic Web (Poster)
Privacy-Preserving Reasoning on the Semantic Web (Poster)Jie Bao
 
Privacy-Preserving Reasoning on the Semantic Web
Privacy-Preserving Reasoning on the Semantic WebPrivacy-Preserving Reasoning on the Semantic Web
Privacy-Preserving Reasoning on the Semantic WebJie Bao
 
Collaborative Construction of Large Biological Ontologies
Collaborative Construction of Large Biological OntologiesCollaborative Construction of Large Biological Ontologies
Collaborative Construction of Large Biological OntologiesJie Bao
 
Representing and Reasoning with Modular Ontologies (2007)
Representing and Reasoning with Modular Ontologies (2007)Representing and Reasoning with Modular Ontologies (2007)
Representing and Reasoning with Modular Ontologies (2007)Jie Bao
 
Query Translation for Ontology-extended Data Sources
Query Translation for Ontology-extended Data SourcesQuery Translation for Ontology-extended Data Sources
Query Translation for Ontology-extended Data SourcesJie Bao
 

Plus de Jie Bao (20)

python-graph-lovestory
python-graph-lovestorypython-graph-lovestory
python-graph-lovestory
 
unix toolbox 中文版
unix toolbox 中文版unix toolbox 中文版
unix toolbox 中文版
 
unixtoolbox.book
unixtoolbox.bookunixtoolbox.book
unixtoolbox.book
 
Lean startup 精益创业 新创企业的成长思维
Lean startup 精益创业 新创企业的成长思维Lean startup 精益创业 新创企业的成长思维
Lean startup 精益创业 新创企业的成长思维
 
Towards social webtops using semantic wiki
Towards social webtops using semantic wikiTowards social webtops using semantic wiki
Towards social webtops using semantic wiki
 
Semantic information theory in 20 minutes
Semantic information theory in 20 minutesSemantic information theory in 20 minutes
Semantic information theory in 20 minutes
 
Towards a theory of semantic communication
Towards a theory of semantic communicationTowards a theory of semantic communication
Towards a theory of semantic communication
 
Owl 2 quick reference card a4 size
Owl 2 quick reference card a4 sizeOwl 2 quick reference card a4 size
Owl 2 quick reference card a4 size
 
ISWC 2010 Metadata Work Summary
ISWC 2010 Metadata Work SummaryISWC 2010 Metadata Work Summary
ISWC 2010 Metadata Work Summary
 
CV
CVCV
CV
 
24 Ways to Explore ISWC 2010 Data
24 Ways to Explore ISWC 2010 Data24 Ways to Explore ISWC 2010 Data
24 Ways to Explore ISWC 2010 Data
 
Semantic Web: In Quest for the Next Generation Killer Apps
Semantic Web: In Quest for the Next Generation Killer AppsSemantic Web: In Quest for the Next Generation Killer Apps
Semantic Web: In Quest for the Next Generation Killer Apps
 
XACML 3.0 (Partial) Concept Map
XACML 3.0 (Partial) Concept MapXACML 3.0 (Partial) Concept Map
XACML 3.0 (Partial) Concept Map
 
Development of a Controlled Natural Language Interface for Semantic MediaWiki
Development of a Controlled Natural Language Interface for Semantic MediaWikiDevelopment of a Controlled Natural Language Interface for Semantic MediaWiki
Development of a Controlled Natural Language Interface for Semantic MediaWiki
 
Digital image self-adaptive acquisition in medical x-ray imaging
Digital image self-adaptive acquisition in medical x-ray imagingDigital image self-adaptive acquisition in medical x-ray imaging
Digital image self-adaptive acquisition in medical x-ray imaging
 
Privacy-Preserving Reasoning on the Semantic Web (Poster)
Privacy-Preserving Reasoning on the Semantic Web (Poster)Privacy-Preserving Reasoning on the Semantic Web (Poster)
Privacy-Preserving Reasoning on the Semantic Web (Poster)
 
Privacy-Preserving Reasoning on the Semantic Web
Privacy-Preserving Reasoning on the Semantic WebPrivacy-Preserving Reasoning on the Semantic Web
Privacy-Preserving Reasoning on the Semantic Web
 
Collaborative Construction of Large Biological Ontologies
Collaborative Construction of Large Biological OntologiesCollaborative Construction of Large Biological Ontologies
Collaborative Construction of Large Biological Ontologies
 
Representing and Reasoning with Modular Ontologies (2007)
Representing and Reasoning with Modular Ontologies (2007)Representing and Reasoning with Modular Ontologies (2007)
Representing and Reasoning with Modular Ontologies (2007)
 
Query Translation for Ontology-extended Data Sources
Query Translation for Ontology-extended Data SourcesQuery Translation for Ontology-extended Data Sources
Query Translation for Ontology-extended Data Sources
 

Extending OWL with Integrity Constraints

  • 1. Extending OWL with Integrity Constraints Jiao Tao1 , Evren Sirin2 , Jie Bao1 , Deborah L. McGuinness1 taoj2@cs.rpi.edu, evren@clarkparsia.com baojie@cs.rpi.edu, dlm@cs.rpi.edu 1 Tetherless World Constellation, Department of Computer Science, Rensselaer Polytechnic Institute, Troy, NY 12180, USA. 2 Clark & Parsia, Washington, DC 20001, USA. The 23rd International Workshop on Description Logics (DL 2010) Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 1 / 20
  • 2. Outline Background OWA and nUNA in OWL IC Requirements for OWL Related Work Contributions Preliminaries SROIQ DCQ IC Semantics for OWL Epistemic Approach Formalization of IC Semantics IC Validation DCQnot From ICs to DCQnot IC Validation Algorithm Implementation Conclusions and Future Work Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 2 / 20
  • 3. Background OWA and nUNA in OWL OWA and nUNA in OWL Open World Assumption (OWA) CWA vs. OWA: any statement that is not known to be true is false vs. unknown Example 1: K satisfies α although p does not have a known producer. K = {Product(p)}, α : Product ∃hasProducer.Producer Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 3 / 20
  • 4. Background OWA and nUNA in OWL OWA and nUNA in OWL Open World Assumption (OWA) CWA vs. OWA: any statement that is not known to be true is false vs. unknown Example 1: K satisfies α although p does not have a known producer. K = {Product(p)}, α : Product ∃hasProducer.Producer nonUnique Name Assumption (nUNA) UNA vs. nUNA: two different names always refer to different entities vs. may refer to the same entity Example 2: K satisfies α although p has two producers. K = {Product(p), hasProducer(p, m1 ), hasProducer(p, m2 )} α : Product ≤ 1hasProducer. . Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 3 / 20
  • 5. Background OWA and nUNA in OWL OWA and nUNA in OWL Open World Assumption (OWA) CWA vs. OWA: any statement that is not known to be true is false vs. unknown Example 1: K satisfies α although p does not have a known producer. K = {Product(p)}, α : Product ∃hasProducer.Producer nonUnique Name Assumption (nUNA) UNA vs. nUNA: two different names always refer to different entities vs. may refer to the same entity Example 2: K satisfies α although p has two producers. K = {Product(p), hasProducer(p, m1 ), hasProducer(p, m2 )} α : Product ≤ 1hasProducer. . Difficult to use OWL for integrity constraint(IC) validation! Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 3 / 20
  • 6. Background IC Requirements for OWL IC Requirements for OWL Strong needs to use OWL as an IC language! Especially in hybrid settings where OWL KBs are integrated with relational databases and ICs are needed to enforce the named individuals in KBs to have some known values. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 4 / 20
  • 7. Background IC Requirements for OWL IC Requirements for OWL Strong needs to use OWL as an IC language! Especially in hybrid settings where OWL KBs are integrated with relational databases and ICs are needed to enforce the named individuals in KBs to have some known values. Typing Constraints Individuals participating in a relation should be instances of certain types, i.e., domain&range axioms and universal restrictions. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 4 / 20
  • 8. Background IC Requirements for OWL IC Requirements for OWL Strong needs to use OWL as an IC language! Especially in hybrid settings where OWL KBs are integrated with relational databases and ICs are needed to enforce the named individuals in KBs to have some known values. Typing Constraints Individuals participating in a relation should be instances of certain types, i.e., domain&range axioms and universal restrictions. Participation Constraints Instances of the constrained class should have a role assertion i.e., existential restrictions. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 4 / 20
  • 9. Background IC Requirements for OWL IC Requirements for OWL Strong needs to use OWL as an IC language! Especially in hybrid settings where OWL KBs are integrated with relational databases and ICs are needed to enforce the named individuals in KBs to have some known values. Typing Constraints Individuals participating in a relation should be instances of certain types, i.e., domain&range axioms and universal restrictions. Participation Constraints Instances of the constrained class should have a role assertion i.e., existential restrictions. Uniqueness Constraints An individual cannot participate in multiple role assertions with the same role, i.e., functional property restrictions, cardinality restrictions. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 4 / 20
  • 10. Background Related Work Related Work Rule-based approach [4, 6] ⊥ ← DL[Product](x), not P(x, y ) P(x, y ) ← DL[hasProducer](x, y ), DL[Producer](y ) IC validation: KDL , ICRule |= ⊥ Limitations: ontology developers need to be familiar with rules Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 5 / 20
  • 11. Background Related Work Related Work Rule-based approach [4, 6] ⊥ ← DL[Product](x), not P(x, y ) P(x, y ) ← DL[hasProducer](x, y ), DL[Producer](y ) IC validation: KDL , ICRule |= ⊥ Limitations: ontology developers need to be familiar with rules Epistemic query-based approach [1] KProduct(x) → ∃y .(KhasProducer(x, y ) ∧ KProducer(y )) IC validation: Ans(ICEQL−Lite , KFOL ) Limitations: complexity results in expressive DLs are unknown Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 5 / 20
  • 12. Background Related Work Related Work Rule-based approach [4, 6] ⊥ ← DL[Product](x), not P(x, y ) P(x, y ) ← DL[hasProducer](x, y ), DL[Producer](y ) IC validation: KDL , ICRule |= ⊥ Limitations: ontology developers need to be familiar with rules Epistemic query-based approach [1] KProduct(x) → ∃y .(KhasProducer(x, y ) ∧ KProducer(y )) IC validation: Ans(ICEQL−Lite , KFOL ) Limitations: complexity results in expressive DLs are unknown Approach based on epistemic extension of DLs [2, 3] KProduct ∃KhasProducer.KProducer IC validation: KDL |= ICEDL Limitations: less expressive DLs, strict UNA Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 5 / 20
  • 13. Background Related Work cont. Related Work cont. Approach reuses OWL as an IC language Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 6 / 20
  • 14. Background Related Work cont. Related Work cont. Approach reuses OWL as an IC language IC validation: T , A |=MM C Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 6 / 20
  • 15. Background Related Work cont. Related Work cont. Approach reuses OWL as an IC language IC validation: T , A |=MM C Limitations: Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 6 / 20
  • 16. Background Related Work cont. Related Work cont. Approach reuses OWL as an IC language IC validation: T , A |=MM C Limitations: Unintuitive Example 4: α is satisfied although the producer of p is unknown. K = {Product(p), ∃hasProducer.Producer(p)} α : Product ∃hasProducer.Producer Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 6 / 20
  • 17. Background Related Work cont. Related Work cont. Approach reuses OWL as an IC language IC validation: T , A |=MM C Limitations: Unintuitive Example 4: α is satisfied although the producer of p is unknown. K = {Product(p), ∃hasProducer.Producer(p)} α : Product ∃hasProducer.Producer Maintenance burden Example 5: α is satisfied although we are not sure if the producer of p is m1 or m2 . K = {Product(p), (∃hasProducer.{m1 , m2 })(p), Producer(m1 ), Producer(m2 ), O(p), O(m1 ), O(m2 )} α : Product ∃hasProducer.(Producer O) Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 6 / 20
  • 18. Background Contributions Contributions The main contributions of this paper include: An IC semantics for OWL A sound and complete IC validation algorithm A prototype implementation Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 7 / 20
  • 19. Preliminaries SROIQ Description Logics SROIQ [5] SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an element of ∆. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 8 / 20
  • 20. Preliminaries SROIQ Description Logics SROIQ [5] SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an element of ∆. I satisfies an axiom α, denoted I |= α, if Type Axiom Condition on I TBox C D C I ⊆ DI R1 R2 I R1 ⊆ R2 I R1 . . . Rn R I ◦ . . . ◦ RI ⊆ RI R1 n RBox Ref(R) ∀x ∈ ∆ : x, x ∈ R I Irr(R) ∀x ∈ ∆ : x, x ∈ R I Dis(R1 , R2 ) I I R1 ∩ R2 = ∅ Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 8 / 20
  • 21. Preliminaries SROIQ Description Logics SROIQ [5] SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an element of ∆. I satisfies an axiom α, denoted I |= α, if Type Axiom Condition on I TBox C D C I ⊆ DI R1 R2 I R1 ⊆ R2 I R1 . . . Rn R I ◦ . . . ◦ RI ⊆ RI R1 n RBox Ref(R) ∀x ∈ ∆ : x, x ∈ R I Irr(R) ∀x ∈ ∆ : x, x ∈ R I Dis(R1 , R2 ) I I R1 ∩ R2 = ∅ I is a model of K if it satisfies all the axioms in K. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 8 / 20
  • 22. Preliminaries SROIQ Description Logics SROIQ [5] SROIQ-interpretation I = (∆, ·I ): ·I maps a concept C to a subset of ∆, a role R to a subset of ∆ × ∆, a named individual NI to an element of ∆. I satisfies an axiom α, denoted I |= α, if Type Axiom Condition on I TBox C D C I ⊆ DI R1 R2 I R1 ⊆ R2 I R1 . . . Rn R I ◦ . . . ◦ RI ⊆ RI R1 n RBox Ref(R) ∀x ∈ ∆ : x, x ∈ R I Irr(R) ∀x ∈ ∆ : x, x ∈ R I Dis(R1 , R2 ) I I R1 ∩ R2 = ∅ I is a model of K if it satisfies all the axioms in K. K entails α, written as K |= α, if I |= α for all models I ∈ Mod(K). Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 8 / 20
  • 23. Preliminaries DCQ Distinguished Conjunctive Query (DCQ) Syntax: Query atom: q ← C (x) | R(x, y ) | ¬R(x, y ) | x = y | x = y Conjunctive Query (CQ): Q ← q | Q1 ∧ Q2 Distinguished Conjunctive Query (DCQ): CQ containing only distinguished variables1 1 A distinguished variable can be mapped to only known individuals, i.e., an element from NI . Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 9 / 20
  • 24. Preliminaries DCQ Distinguished Conjunctive Query (DCQ) Syntax: Query atom: q ← C (x) | R(x, y ) | ¬R(x, y ) | x = y | x = y Conjunctive Query (CQ): Q ← q | Q1 ∧ Q2 Distinguished Conjunctive Query (DCQ): CQ containing only distinguished variables1 Semantics: Given a KB K, a DCQ Q, and an assignment σ : NV → NI , we say that K |=σ Q if: K |=σ q iff K |= σ(q) σ K |= Q1 ∧ Q2 iff K |=σ Q1 and K |=σ Q2 where σ(Q) denotes the application of σ to Q. Ans(Q, K) = {σ | K |=σ Q} K |= Q if Ans(Q, K) = ∅ 1 A distinguished variable can be mapped to only known individuals, i.e., an element from NI . Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 9 / 20
  • 25. IC Semantics for OWL Epistemic Approach Epistemic Approach ICs are epistemic in nature [8]: about what the KB knows. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 10 / 20
  • 26. IC Semantics for OWL Epistemic Approach Epistemic Approach ICs are epistemic in nature [8]: about what the KB knows. We adopt this epistemic view of ICs and propose an IC semantics for OWL, which is similar to how the semantics of epistemic DL ALCK [2] and MKNF DL ALCKN F [3] are defined, but is still different from them: Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 10 / 20
  • 27. IC Semantics for OWL Epistemic Approach Epistemic Approach ICs are epistemic in nature [8]: about what the KB knows. We adopt this epistemic view of ICs and propose an IC semantics for OWL, which is similar to how the semantics of epistemic DL ALCK [2] and MKNF DL ALCKN F [3] are defined, but is still different from them: weak UNA vs. strict UNA expressive DLs, i.e., SROIQ, vs. less expressive DLs, i.e., ALC. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 10 / 20
  • 28. IC Semantics for OWL Formalization of IC Semantics IC-interpretation An IC-interpretation is a pair I, U where I is a SROIQ interpretation defined over ∆I and U is a set of SROIQ interpretations. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 11 / 20
  • 29. IC Semantics for OWL Formalization of IC Semantics IC-interpretation An IC-interpretation is a pair I, U where I is a SROIQ interpretation defined over ∆I and U is a set of SROIQ interpretations. IC-interpretations of concepts, roles, individuals: C I,U = {x I | x ∈ NI s.t. ∀J ∈ U, x J ∈ C J } R I,U = { x I , y I | x, y ∈ NI s.t. ∀J ∈ U, x J , y J ∈ R J } (R − )I,U = { x I , y I | y I , x I ∈ R I,U } (C D)I,U = C I,U ∩ D I,U , (¬C )I,U = NI C I,U (≥ nR.C )I,U = {x I | x ∈ NI s.t. #{y I | x I , y I ∈ R I,U , y I ∈ C I,U } ≥ n} (∃R.Self)I,U = {x I | x ∈ NI s.t. x I , x I ∈ R I,U }, {a}I,U = {aI }. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 11 / 20
  • 30. IC Semantics for OWL Formalization of IC Semantics IC-interpretation An IC-interpretation is a pair I, U where I is a SROIQ interpretation defined over ∆I and U is a set of SROIQ interpretations. IC-interpretations of concepts, roles, individuals: C I,U = {x I | x ∈ NI s.t. ∀J ∈ U, x J ∈ C J } R I,U = { x I , y I | x, y ∈ NI s.t. ∀J ∈ U, x J , y J ∈ R J } (R − )I,U = { x I , y I | y I , x I ∈ R I,U } (C D)I,U = C I,U ∩ D I,U , (¬C )I,U = NI C I,U (≥ nR.C )I,U = {x I | x ∈ NI s.t. #{y I | x I , y I ∈ R I,U , y I ∈ C I,U } ≥ n} (∃R.Self)I,U = {x I | x ∈ NI s.t. x I , x I ∈ R I,U }, {a}I,U = {aI }. IC-interpretations use the CWA w.r.t. NI (standard DL interpretations: OWA). Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 11 / 20
  • 31. IC Semantics for OWL Formalization of IC Semantics Weak UNA Weak UNA: two named individuals with different identifiers are assumed to be different by default unless their equality is required to satisfy the axioms in the KB. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 12 / 20
  • 32. IC Semantics for OWL Formalization of IC Semantics Weak UNA Weak UNA: two named individuals with different identifiers are assumed to be different by default unless their equality is required to satisfy the axioms in the KB. How weak UNA is formalized? Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 12 / 20
  • 33. IC Semantics for OWL Formalization of IC Semantics Weak UNA Weak UNA: two named individuals with different identifiers are assumed to be different by default unless their equality is required to satisfy the axioms in the KB. How weak UNA is formalized? J = I if: For every concept C, J |= C (a) implies I |= C (a); For every role R, J |= R(a, b) implies I |= R(a, b); EJ ⊂ EI where EI = { a, b | a, b ∈ NI s.t. I |= a = b}. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 12 / 20
  • 34. IC Semantics for OWL Formalization of IC Semantics Weak UNA Weak UNA: two named individuals with different identifiers are assumed to be different by default unless their equality is required to satisfy the axioms in the KB. How weak UNA is formalized? J = I if: For every concept C, J |= C (a) implies I |= C (a); For every role R, J |= R(a, b) implies I |= R(a, b); EJ ⊂ EI where EI = { a, b | a, b ∈ NI s.t. I |= a = b}. ME models: the models of K with minimal equality (ME) between NI : ModME (K) = {I ∈ Mod(K) | J , J ∈ Mod(K), J = I} Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 12 / 20
  • 35. IC Semantics for OWL Formalization of IC Semantics Weak UNA Weak UNA: two named individuals with different identifiers are assumed to be different by default unless their equality is required to satisfy the axioms in the KB. How weak UNA is formalized? J = I if: For every concept C, J |= C (a) implies I |= C (a); For every role R, J |= R(a, b) implies I |= R(a, b); EJ ⊂ EI where EI = { a, b | a, b ∈ NI s.t. I |= a = b}. ME models: the models of K with minimal equality (ME) between NI : ModME (K) = {I ∈ Mod(K) | J , J ∈ Mod(K), J = I} IC-semantics uses the weak UNA (standard DL semantics: nUNA). Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 12 / 20
  • 36. IC Semantics for OWL Formalization of IC Semantics Axiom Satisfaction, IC Satisfaction, Extended KB An IC-interpretation I, U satisfies an axiom α, denoted as I, U |= α,if Type Axiom Condition on I, U TBox C D C I,U ⊆ D I,U I,U I,U R1 R2 R1 ⊆ R2 I,U I,U R1 . . . Rn R R1 ◦ . . . ◦ Rn ⊆ R I,U RBox Ref(R) ∀x ∈ NI : x I,U , x I,U ∈ R I,U Irr(R) ∀x ∈ NI : x I,U , x I,U ∈ R I,U I,U I,U Dis(R1 , R2 ) R1 ∩ R2 = ∅ Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 13 / 20
  • 37. IC Semantics for OWL Formalization of IC Semantics Axiom Satisfaction, IC Satisfaction, Extended KB An IC-interpretation I, U satisfies an axiom α, denoted as I, U |= α,if Type Axiom Condition on I, U TBox C D C I,U ⊆ D I,U I,U I,U R1 R2 R1 ⊆ R2 I,U I,U R1 . . . Rn R R1 ◦ . . . ◦ Rn ⊆ R I,U RBox Ref(R) ∀x ∈ NI : x I,U , x I,U ∈ R I,U Irr(R) ∀x ∈ NI : x I,U , x I,U ∈ R I,U I,U I,U Dis(R1 , R2 ) R1 ∩ R2 = ∅ The IC-satisfaction of α by K, i.e., K |=IC α, is defined as: K |=IC α iff ∀I ∈ U, I, U |= α, where U = ModME (K) Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 13 / 20
  • 38. IC Semantics for OWL Formalization of IC Semantics Axiom Satisfaction, IC Satisfaction, Extended KB An IC-interpretation I, U satisfies an axiom α, denoted as I, U |= α,if Type Axiom Condition on I, U TBox C D C I,U ⊆ D I,U I,U I,U R1 R2 R1 ⊆ R2 I,U I,U R1 . . . Rn R R1 ◦ . . . ◦ Rn ⊆ R I,U RBox Ref(R) ∀x ∈ NI : x I,U , x I,U ∈ R I,U Irr(R) ∀x ∈ NI : x I,U , x I,U ∈ R I,U I,U I,U Dis(R1 , R2 ) R1 ∩ R2 = ∅ The IC-satisfaction of α by K, i.e., K |=IC α, is defined as: K |=IC α iff ∀I ∈ U, I, U |= α, where U = ModME (K) An extended KB is a pair K, C where K is a SROIQ KB interpreted with the standard OWL semantics and C is a set of SROIQ axioms interpreted with IC semantics. We say that K, C is valid if ∀α ∈ C, K |=IC α. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 13 / 20
  • 39. IC Validation DCQnot Distinguished Conjunctive Query with not (DCQnot ) DCQnot is the extension of DCQ with the NAF operator not . Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 14 / 20
  • 40. IC Validation DCQnot Distinguished Conjunctive Query with not (DCQnot ) DCQnot is the extension of DCQ with the NAF operator not . Syntax: Q ← q | Q1 ∧ Q2 | not Q | Q1 ∨ Q2 where Q1 ∨ Q2 is the abbreviation for not (not Q1 ∧ not Q2 ). Semantics: K |=σ not Q iff σ s.t. K |=σ σ(Q) K |=σ Q1 ∨ Q2 iff K |=σ Q1 or K |=σ Q2 Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 14 / 20
  • 41. IC Validation From ICs to DCQnot From ICs to DCQnot Translation from ICs to DCQnot : translate an IC axiom α into a query Q such that when α is satisfied, Q is false. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 15 / 20
  • 42. IC Validation From ICs to DCQnot From ICs to DCQnot Translation from ICs to DCQnot : translate an IC axiom α into a query Q such that when α is satisfied, Q is false. Translation rules: Concept Translation: Tc (Ca , x) := Ca (x) Tc (¬C , x) := not Tc (C , x) Tc (C1 C2 , x) := Tc (C1 , x) ∧ Tc (C2 , x) Tc (≥ nR.C , x) := (R(x, yi ) ∧ Tc (C , yi )) not (yi = yj ) 1≤i≤n 1≤i<j≤n Tc (∃R.Self, x) := R(x, x) Tc ({a}, x) := (x = a) Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 15 / 20
  • 43. IC Validation From ICs to DCQnot cont. From ICs to DCQnot cont. Translation rules cont.: Axiom Translation: T (C1 C2 ) := Tc (C1 , x) ∧ not Tc (C2 , x) T (R1 R2 ) := R1 (x, y ) ∧ not R2 (x, y ) T (R1 . . . Rn R) := R1 (x, y1 ) ∧ . . . Rn (yn−1 , yn ) ∧ not R(x, yn ) T (Ref(R)) := not R(x, x) T (Irr(R)) := R(x, x) T (Dis(R1 , R2 )) := R1 (x, y ) ∧ R2 (x, y ) Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 16 / 20
  • 44. IC Validation From ICs to DCQnot cont. From ICs to DCQnot cont. Translation rules cont.: Axiom Translation: T (C1 C2 ) := Tc (C1 , x) ∧ not Tc (C2 , x) T (R1 R2 ) := R1 (x, y ) ∧ not R2 (x, y ) T (R1 . . . Rn R) := R1 (x, y1 ) ∧ . . . Rn (yn−1 , yn ) ∧ not R(x, yn ) T (Ref(R)) := not R(x, x) T (Irr(R)) := R(x, x) T (Dis(R1 , R2 )) := R1 (x, y ) ∧ R2 (x, y ) Example 6: T (Product ∃hasProducer.Producer) := Tc (Product, x) ∧ not Tc (∃hasProducer.Producer, x) := Product(x) ∧ not (hasProducer(x, y ) ∧ Tc (Producer, y )) := Product(x) ∧ not (hasProducer(x, y ) ∧ Producer(y )) Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 16 / 20
  • 45. IC Validation IC Validation Algorithm IC Validation Algorithm Theorem: Given an extended KB K, C with expressivity SRI, SROIQ ( SROIQ, SROI resp.), we have that K |=IC α iff K |= T (α) where α ∈ C [9]. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 17 / 20
  • 46. IC Validation IC Validation Algorithm IC Validation Algorithm Theorem: Given an extended KB K, C with expressivity SRI, SROIQ ( SROIQ, SROI resp.), we have that K |=IC α iff K |= T (α) where α ∈ C [9]. To limit the interactions between the disjunctive (in)equality axioms in K and the cardinality restrictions in C, we can require: Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 17 / 20
  • 47. IC Validation IC Validation Algorithm IC Validation Algorithm Theorem: Given an extended KB K, C with expressivity SRI, SROIQ ( SROIQ, SROI resp.), we have that K |=IC α iff K |= T (α) where α ∈ C [9]. To limit the interactions between the disjunctive (in)equality axioms in K and the cardinality restrictions in C, we can require: K: within SRI, no nominals & cardinality restrictions thus no disjunctive (in)equality; C: within SROI, no cardinality restrictions. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 17 / 20
  • 48. Implementation Implementation Prototype: IC validator 2 in the OWL 2 DL reasoner Pellet. SPARQL-based [7] implementation. 2 http://clarkparsia.com/pellet/oicv-0.1.2.zip Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 18 / 20
  • 49. Implementation Implementation Prototype: IC validator 2 in the OWL 2 DL reasoner Pellet. SPARQL-based [7] implementation. The validation process is as follows: Reading OWL IC axioms Translating to DCQnot Translating to SPARQL query Executed by the SPARQL query engine 2 http://clarkparsia.com/pellet/oicv-0.1.2.zip Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 18 / 20
  • 50. Implementation Implementation Prototype: IC validator 2 in the OWL 2 DL reasoner Pellet. SPARQL-based [7] implementation. The validation process is as follows: Reading OWL IC axioms Translating to DCQnot Translating to SPARQL query Executed by the SPARQL query engine Preliminary evaluation: Dataset: LUBM(5), 100K individuals, 800K ABox axioms IC validation time: 2s Logical consistency checking time: 10s 2 http://clarkparsia.com/pellet/oicv-0.1.2.zip Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 18 / 20
  • 51. Conclusions and Future Work Conclusions and Future Work In this paper, we have described an IC semantics for OWL presented a sound and complete IC validation algorithm described the prototype IC validator Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 19 / 20
  • 52. Conclusions and Future Work Conclusions and Future Work In this paper, we have described an IC semantics for OWL presented a sound and complete IC validation algorithm described the prototype IC validator In the future, we will: look at the performance of IC validation in real-world datasets explore the IC validation algorithms for the fullly expressive DLs Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 19 / 20
  • 53. Questions Questions? Thanks for your attention! Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 20 / 20
  • 54. Questions D. Calvanese, G. D. Giacomo, D. Lembo, M. Lenzerini, and R. Rosati. EQL-Lite: Effective First-Order Query Processing in Description Logics. In Proc. IJCAI2007, pages 274–279, 2007. F. M. Donini, M. Lenzerini, D. Nardi, W. Nutt, and A. Schaerf. An Epistemic Operator for Description Logics. AI, 100(1–2):225–274, 1998. F. M. Donini, D. Nardi, and R. Rosati. Description Logics of Minimal Knowledge and Negation as Failure. ACM Trans. Comput. Logic, 3(2):177–225, 2002. T. Eiter, G. Ianni, T. Lukasiewicz, R. Schindlauer, and H. Tompits. Combining Answer Set Programming with Description Logics for the Semantic Web. AI, 172(12-13):1495–1539, 2008. I. Horrocks, O. Kutz, and U. Sattler. The Even More Irresistible SROIQ. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 20 / 20
  • 55. Questions In Proc. KR2006, pages 57–67, 2006. B. Motik. A Faithful Integration of Description Logics with Logic Programming. In Proc. IJCAI2007, pages 477–482, 2007. E. Prud’hommeaux and A. Seaborne. SPARQL Query Language for RDF, 2008. R. Reiter. On Integrity Constraints. In Proc. TARK1988, pages 97–111, 1988. J. Tao, E. Sirin, J. Bao, and D. McGuinness. Integrity Constraints in OWL. Technical report, Dept. of Computer Science, Rensselaer Polytechnic Institute, Troy, NY., 2010. Tao,Sirin,Bao,McGuinness (RPI,Clark Parsia) Extending OWL with Integrity Constraints May 2010, DL 2010 20 / 20