Reasoning with DAML+OIL:

Transcription

1 Reasoning with DAML+OIL: What can it do for YOU? Ian Horrocks University of Manchester Manchester, UK DAML PI meeting, Nashua, July 2001 p.1/9

2 DAML+OIL Language Overview DAML+OIL is an ontology language DAML PI meeting, Nashua, July 2001 p.2/9

3 DAML+OIL Language Overview DAML+OIL is an ontology language Describes structure of the domain (i.e., a schema) RDF used to describe specific instance of domain (data) DAML PI meeting, Nashua, July 2001 p.2/9

4 DAML+OIL Language Overview DAML+OIL is an ontology language Describes structure of the domain (i.e., a schema) RDF used to describe specific instance of domain (data) Structure described in terms of classes and properties DAML PI meeting, Nashua, July 2001 p.2/9

5 DAML+OIL Language Overview DAML+OIL is an ontology language Describes structure of the domain (i.e., a schema) RDF used to describe specific instance of domain (data) Structure described in terms of classes and properties Ontology consists of set of axioms E.g., asserting class subsumption/equivalence DAML PI meeting, Nashua, July 2001 p.2/9

6 DAML+OIL Language Overview DAML+OIL is an ontology language Describes structure of the domain (i.e., a schema) RDF used to describe specific instance of domain (data) Structure described in terms of classes and properties Ontology consists of set of axioms E.g., asserting class subsumption/equivalence Classes can be names or expressions Various constructors provided for building class expressions DAML PI meeting, Nashua, July 2001 p.2/9

7 DAML+OIL Language Overview DAML+OIL is an ontology language Describes structure of the domain (i.e., a schema) RDF used to describe specific instance of domain (data) Structure described in terms of classes and properties Ontology consists of set of axioms E.g., asserting class subsumption/equivalence Classes can be names or expressions Various constructors provided for building class expressions Expressive power determined by Kinds of axiom supported Kinds of class (and property) constructor supported DAML PI meeting, Nashua, July 2001 p.2/9

8 DAML+OIL Class Constructors Constructor Abbreviation Example intersectionof C 1... C n Human Male unionof C 1... C n Doctor Lawyer complementof C Male oneof {x 1... x n } {john, mary} toclass P.C haschild.doctor hasclass P.C haschild.lawyer hasvalue P.{x} citizenof.{usa} mincardinalityq n P.C 2 haschild.lawyer maxcardinalityq n P.C 1 haschild.male cardinalityq =n P.C =1 hasparent.female DAML PI meeting, Nashua, July 2001 p.3/9

9 DAML+OIL Class Constructors Constructor Abbreviation Example intersectionof C 1... C n Human Male unionof C 1... C n Doctor Lawyer complementof C Male oneof {x 1... x n } {john, mary} toclass P.C haschild.doctor hasclass P.C haschild.lawyer hasvalue P.{x} citizenof.{usa} mincardinalityq n P.C 2 haschild.lawyer maxcardinalityq n P.C 1 haschild.male cardinalityq =n P.C =1 hasparent.female Arbitrarily complex nesting of constructors E.g., haschild.(doctor haschild.doctor) DAML PI meeting, Nashua, July 2001 p.3/9

10 DAML+OIL Class Constructors Constructor Abbreviation Example intersectionof C 1... C n Human Male unionof C 1... C n Doctor Lawyer complementof C Male oneof {x 1... x n } {john, mary} toclass P.C haschild.doctor hasclass P.C haschild.lawyer hasvalue P.{x} citizenof.{usa} mincardinalityq n P.C 2 haschild.lawyer maxcardinalityq n P.C 1 haschild.male cardinalityq =n P.C =1 hasparent.female Arbitrarily complex nesting of constructors E.g., haschild.(doctor haschild.doctor) XMLS datatypes as well as classes DAML PI meeting, Nashua, July 2001 p.3/9

11 DAML+OIL Axioms Axiom Abbreviation Example subclassof C 1 C 2 Human Animal Biped. sameclassas C 1 = C2 Man =. Human Male subpropertyof P 1 P 2 hasdaughter haschild. samepropertyas P 1 = P2 cost =. price. sameindividualas x 1 = x2 President_Bush =. G_W_Bush disjointwith C 1 C 2 Male Female differentindividualfrom {x 1 } {x 2 } {john} {peter}. inverseof P 1 = P 2 haschild =. hasparent transitiveproperty P + P ancestor + ancestor uniqueproperty Thing 1P Thing 1hasMother UnambiguousProperty Thing 1P Thing 1isMotherOf DAML PI meeting, Nashua, July 2001 p.4/9

12 DAML+OIL Axioms Axiom Abbreviation Example subclassof C 1 C 2 Human Animal Biped. sameclassas C 1 = C2 Man =. Human Male subpropertyof P 1 P 2 hasdaughter haschild. samepropertyas P 1 = P2 cost =. price. sameindividualas x 1 = x2 President_Bush =. G_W_Bush disjointwith C 1 C 2 Male Female differentindividualfrom {x 1 } {x 2 } {john} {peter}. inverseof P 1 = P 2 haschild =. hasparent transitiveproperty P + P ancestor + ancestor uniqueproperty Thing 1P Thing 1hasMother UnambiguousProperty Thing 1P Thing 1isMotherOf Axioms (mostly) reducible to subclass/propertyof DAML PI meeting, Nashua, July 2001 p.4/9

13 Decidable Reasoning Set of operators/axioms restricted so that reasoning is decidable DAML PI meeting, Nashua, July 2001 p.5/9

14 Decidable Reasoning Set of operators/axioms restricted so that reasoning is decidable Significant point on tractability -v- expressiveness scale DAML PI meeting, Nashua, July 2001 p.5/9

15 Decidable Reasoning Set of operators/axioms restricted so that reasoning is decidable Significant point on tractability -v- expressiveness scale Consistent with Semantic Web s layered architecture XML provides syntax transport layer RDF provides basic ontological primitives DAML+OIL provides (decidable) logical layer Further layers (e.g., rules) will extend DAML+OIL DAML PI meeting, Nashua, July 2001 p.5/9

16 Decidable Reasoning Set of operators/axioms restricted so that reasoning is decidable Significant point on tractability -v- expressiveness scale Consistent with Semantic Web s layered architecture XML provides syntax transport layer RDF provides basic ontological primitives DAML+OIL provides (decidable) logical layer Further layers (e.g., rules) will extend DAML+OIL Facilitates provision of reasoning services Known algorithms Implemented systems Evidence of empirical tractability DAML PI meeting, Nashua, July 2001 p.5/9

17 Why Reasoning Services? Reasoning is important for: DAML PI meeting, Nashua, July 2001 p.6/9

18 Why Reasoning Services? Reasoning is important for: Ontology design Check class consistency and (unexpected) implied relationships Particularly important with large ontologies/multiple authors DAML PI meeting, Nashua, July 2001 p.6/9

19 Why Reasoning Services? Reasoning is important for: Ontology design Check class consistency and (unexpected) implied relationships Particularly important with large ontologies/multiple authors Ontology integration Assert inter-ontology relationships Reasoner computes integrated class hierarchy/consistency DAML PI meeting, Nashua, July 2001 p.6/9

20 Why Reasoning Services? Reasoning is important for: Ontology design Check class consistency and (unexpected) implied relationships Particularly important with large ontologies/multiple authors Ontology integration Assert inter-ontology relationships Reasoner computes integrated class hierarchy/consistency Ontology deployment Determine if set of facts are consistent w.r.t. ontology Determine if individuals are instances of ontology classes No point in having semantics unless exploited by agents DAML PI meeting, Nashua, July 2001 p.6/9

21 Why Reasoning Services? Reasoning is important for: Ontology design Check class consistency and (unexpected) implied relationships Particularly important with large ontologies/multiple authors Ontology integration Assert inter-ontology relationships Reasoner computes integrated class hierarchy/consistency Ontology deployment Determine if set of facts are consistent w.r.t. ontology Determine if individuals are instances of ontology classes No point in having semantics unless exploited by agents The Semantic Web needs a logic on top (Henry Thompson) DAML PI meeting, Nashua, July 2001 p.6/9

22 OilEd OilEd is a DAML+OIL ontology editor with reasoning support DAML PI meeting, Nashua, July 2001 p.7/9

23 OilEd OilEd is a DAML+OIL ontology editor with reasoning support Frame based interface (inspired by Protegé) DAML PI meeting, Nashua, July 2001 p.7/9

24 OilEd OilEd is a DAML+OIL ontology editor with reasoning support Frame based interface (inspired by Protegé) Extended to clarify semantics and capture whole language Explicit (hasclass) or (toclass) restrictions Boolean connectives (,, ) and nesting Transitive and unique (functional) properties DAML PI meeting, Nashua, July 2001 p.7/9

25 OilEd OilEd is a DAML+OIL ontology editor with reasoning support Frame based interface (inspired by Protegé) Extended to clarify semantics and capture whole language Explicit (hasclass) or (toclass) restrictions Boolean connectives (,, ) and nesting Transitive and unique (functional) properties Reasoning support provided by FaCT system Ontology translated into SHIQ DL Communicates with FaCT via CORBA interface Indicates inconsistencies and implicit subsumptions Can add axioms to make implicit subsumptions explicit DAML PI meeting, Nashua, July 2001 p.7/9

26 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology DAML PI meeting, Nashua, July 2001 p.8/9

27 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. DAML PI meeting, Nashua, July 2001 p.8/9

28 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death DAML PI meeting, Nashua, July 2001 p.8/9

29 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death Multiple equality/inclusion axioms DAML PI meeting, Nashua, July 2001 p.8/9

30 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death Multiple equality/inclusion axioms Stomach-Ulcer. = Ulcer haslocation.stomach plus Stomach-Ulcer haslocation.lining-of-stomach Ulcer haslocation.stomach OrganLiningLesion DAML PI meeting, Nashua, July 2001 p.8/9

31 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death Multiple equality/inclusion axioms Stomach-Ulcer. = Ulcer haslocation.stomach plus Stomach-Ulcer haslocation.lining-of-stomach Ulcer haslocation.stomach OrganLiningLesion Inverse roles capture e.g. causes/causedby relationship DAML PI meeting, Nashua, July 2001 p.8/9

32 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death Multiple equality/inclusion axioms Stomach-Ulcer. = Ulcer haslocation.stomach plus Stomach-Ulcer haslocation.lining-of-stomach Ulcer haslocation.stomach OrganLiningLesion Inverse roles capture e.g. causes/causedby relationship Death causedby.smoking PrematureDeath Smoking causes.prematuredeath DAML PI meeting, Nashua, July 2001 p.8/9

33 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death Multiple equality/inclusion axioms Stomach-Ulcer. = Ulcer haslocation.stomach plus Stomach-Ulcer haslocation.lining-of-stomach Ulcer haslocation.stomach OrganLiningLesion Inverse roles capture e.g. causes/causedby relationship Death causedby.smoking PrematureDeath Smoking causes.prematuredeath Cardinality restrictions add consistency constraints DAML PI meeting, Nashua, July 2001 p.8/9

34 Reasoning Examples what you CAN do E.g., DAML+OIL medical terminology ontology Transitive roles capture partonomy, causality, etc. Smoking causes.cancer plus Cancer causes.death Smoking causes.death Multiple equality/inclusion axioms Stomach-Ulcer. = Ulcer haslocation.stomach plus Stomach-Ulcer haslocation.lining-of-stomach Ulcer haslocation.stomach OrganLiningLesion Inverse roles capture e.g. causes/causedby relationship Death causedby.smoking PrematureDeath Smoking causes.prematuredeath Cardinality restrictions add consistency constraints BloodPressure hasvalue.(high Low) 1hasValue plus High Low HighLowBloodPressure DAML PI meeting, Nashua, July 2001 p.8/9

35 Reasoning Examples what you CAN T do Where to begin! DAML PI meeting, Nashua, July 2001 p.9/9

36 Reasoning Examples what you CAN T do Where to begin! Robust decidability largely due to tree model property For any consistent class there exists a tree (like) model DAML PI meeting, Nashua, July 2001 p.9/9

37 Reasoning Examples what you CAN T do Where to begin! Robust decidability largely due to tree model property For any consistent class there exists a tree (like) model No property constructors, e.g.: parent brother uncle ancestor. = parent + DAML PI meeting, Nashua, July 2001 p.9/9

38 Reasoning Examples what you CAN T do Where to begin! Robust decidability largely due to tree model property For any consistent class there exists a tree (like) model No property constructors, e.g.: parent brother uncle ancestor. = parent + No variables, e.g.: Ulcer haslocation.?x haslocation.( LiningOf.?x) DAML PI meeting, Nashua, July 2001 p.9/9

39 Reasoning Examples what you CAN T do Where to begin! Robust decidability largely due to tree model property For any consistent class there exists a tree (like) model No property constructors, e.g.: parent brother uncle ancestor. = parent + No variables, e.g.: Ulcer haslocation.?x haslocation.( LiningOf.?x) Only have unary and binary predicates Can t express (directly) P (x, y, z) DAML PI meeting, Nashua, July 2001 p.9/9

40 Reasoning Examples what you CAN T do Where to begin! Robust decidability largely due to tree model property For any consistent class there exists a tree (like) model No property constructors, e.g.: parent brother uncle ancestor. = parent + No variables, e.g.: Ulcer haslocation.?x haslocation.( LiningOf.?x) Only have unary and binary predicates Can t express (directly) P (x, y, z) Language extensions may remove some of above limitations DAML PI meeting, Nashua, July 2001 p.9/9

41 Reasoning Examples what you CAN T do Where to begin! Robust decidability largely due to tree model property For any consistent class there exists a tree (like) model No property constructors, e.g.: parent brother uncle ancestor. = parent + No variables, e.g.: Ulcer haslocation.?x haslocation.( LiningOf.?x) Only have unary and binary predicates Can t express (directly) P (x, y, z) Language extensions may remove some of above limitations But there is no such thing as a free lunch DAML PI meeting, Nashua, July 2001 p.9/9