320 likes | 467 Views
AMRUTA PRIYA MOGILINEEDI apm075000@utdallas.edu. OWL 2 Web Ontology Language. Topics. Introduction to OWL Usage of OWL Problems with OWL 1 Solutions from OWL 2. OWL Introduction.
E N D
AMRUTA PRIYA MOGILINEEDI apm075000@utdallas.edu OWL 2 Web Ontology Language
Topics • Introduction to OWL • Usage of OWL • Problems with OWL 1 • Solutions from OWL 2
OWL Introduction • Initially focus was on RDF and RDF Schema • OWL is a family of three language variants 1. OWL Lite 2. OWL DL 3. OWL Full • OWL is a standardized language.
Usage of OWL • OWL Ontologies are used in areas such as e-Science, medicine, biology, geography, astronomy, defense, automotive and aerospace industries. • OWL is extensively used in life sciences community.
Problems in OWL 1 • Expressivity Issues • Problems with syntax • Deficiencies in the definition
Expressivity Limitations • Lacks several constructs • Community users introduced various patterns • Still Incomplete • Problem with 1) Qualified Cardinality Restriction 2) Relational Expressivity 3) Data type Expressivity 4) Keys
1) Qualified Cardinality Restriction(QCR) • Ex: “Persons that have at least one child who is male” ---existential restriction (qualified) • Ex: “Person with at least 3 children who are male”---cardinality restriction(not qualified) • Some situations where QCR is used i) Define a quadruped animal with exactly 4 parts that are legs. ii) Medical Committee consisting of at least 5 members, 2 are medically qualified, 1 is manager and 2 are public members • QCR was used in DAML + OIL---predecessor of OWL
2)Relational Expressivity • Propagation along properties Ex: “Abnormality of a part of an anatomical structure constitutes an abnormality of the structure as a whole”. • Properties of Properties Ex: “If x is part of y and y is part of z then x is part of z” reflexive(every object is a part of itself) asymmetric (nothing is a part of one of its parts).
Data type Expressivity • “British citizen must have a passport number which is an xsd: string” • We cannot express 1) restrictions to a subset of datatype values 2) relationships between values of data properties on one object 3) relationships between values of data properties on different objects 4)aggregate functions
4) Keys • OWL DL does not support Key Constraints on data properties. Ex: “US Citizens are uniquely identified by their SSN” This is not possible • OWL Full supports this using inverse functional properties.
2) Syntax Issues • OWL 1 uses 2 types of syntax 1. Abstract Syntax 2. OWL 1 RDF • Both are difficult to parse.
1) Frame-Based Paradigm • Axiom Ex: “Tiger is a Class and it is a subclass of Class Cat” Class(Tiger partial Cat) “Class Tiger is a subclass of Class Predator” Adding, SubClassOf(Tiger Predator) Ans: Cat and Predator
2) Alignment with DL Constructs • DL allows only one class to appear in property restriction Ex: restriction( hasParent SomeValuesFrom(Person) allValuesFrom(Person)) This can be translated into intersectionOf( restriction(hasParent SomeValuesFrom(Person)) restriction(hasParent allValuesFrom(Person))
3) Types of Ontology Entities • Ex: class(Person partial restriction(hasMother SomeValuesFrom(Woman)) • Ex: DisjointClasses(Animal Plant) SubClassOf(Human Animla)
4) Problems with OWL 1 RDF • Information is represented in triples • Ex: A is UNION of B and C (A, owl:unionOf, _:x1) (_:x1, rdf:first, B) (_:x1, rdf:rest, _:x2) (_:x2, rdf:first, C) (_:x2, rdf:rest, rdf:nil) This is difficult to read
Metamodeling • Ex: “Harry is an Eagle” “Eagles are endangered species” --Individual Harry is an instance of Class Eagle --Individual Eagle id an instance of Class Endangered Species. • Eagle is Individual in one and Class in the another.
Imports and Versioning • Imports one ontology from another • URI must point to the location of imported ontology and the location should match with the name of imported ontology. • Coupling of names and location is not possible when ontologies are moved.
Solutions from OWL 21)Qualified Number Restrictions • QCR is implemented in OWL 2
2) Relational Expressivity • Addressed using Complex Property Inclusion Axioms Ex: “If a contains b and b has a part c, then a also contains c” subPropertyOf( Propertychain(contains hasPart) contains) Ex: “If b is part of a and b contains c , then c is also a part of a” subProperty( Propertychain(hasParthasPart) hasPart) ---Cycle of dependency between contains and hasPart • OWL 2 supports transitive, reflexive and asymmetric.
3) Increasing Data type Expressivity • OWL 2 supports owl:boolean, owl:string, xsd:integer, xsd:dateTime, xsd:hexBinary • Build new data types by providing data type restriction construct. Ex: DatatypeRestriction(xsd:integer xsd:minInclusive 18)
4) Keys • OWL 2 implements Easy Keys. • Key Axioms are of the form HashKey(C P1,……Pn) This states that object or data type properties Pi are keys for instance of Class C. • No two instances of C can coincide on values of Pi. Ex: HashKey(Person hasSSN)
2 assertions can be made PropertyAssertion(PSmithhasSSN “123-45-6789”) PropertyAssertion(PeterSmithhasSSN “123-45-6789”) ---Psmith and PeterSmith are same individuals. • Easy Keys are not applied to individuals not known by name Ex: Jane is connected through marriedTo to an individual x that is an instance of Man and that has “123-45-6789” as the value of hasSSN.
ClassAssertion( SomeValuesFrom(marriedTo IntersectionOf( Man HasValue(hasSSN “123-45-6789”))) Jane) --We cannot conclude that ClassAssertion(Man Psmith)
MOF Metamodel • It is a meta language and used for specifying other languages. • It is also a Structural Specification presented using UML.
Typing and Declarations • Ex: Declaration(Class(Plant)) Declaration(Class(Animal)) • SubPropertyOf(P Q) Both P and Q must be declared, otherwise we cannot conclude whether they are objects or data type properties.
Metamodeling with Punning • Ex: “Eagles are endangered species” “Harry is an Eagle” ------ClassAssertion(Eagle Harry) ClassAssertion(Endangered Eagle) This can be interpreted in OWL 2 as ClassAssertion(Class-Eagle Individual-Harry) ClassAssertion(Class-Endangered Individual-Eagle) ---This type of meta modeling is referred as Punning.
Annotation of Axiom • Ex: “Humans are animals and attributes that statement to Peter” SubClassOf( Annotation(attributedTo “Peter”) Human Animal) ---Annotations do not affect the semantics of ontology but they do affect the structural equivalence. • SubClassOf(Human Animal) This is semantically equivalent to the above one but not structurally.
Imports and Versioning • OWL 2 allows redirection mechanism. • OWL 2 contains Version URI which identifies the version of the ontology.
Conclusion • Apart from addressing problems in OWL 1, goal of OWL 2 is to provide a robust platform for future development.