Semantic Web and The Web Ontology Language

1. Semantic Web and The Web Ontology Language

2. The World Wide Web today • Information is represented • in natural language • in multiple formats • using graphics, multimedia • Information overload • Though humans can process this information, • machines (software agents) cannot understand it • vast majority of information remains untapped

3. Semantic Web Goals • Make information on the Web machine processable • Provide a standard data model for data integration • Build machine reasoning infrastructure • Enhance the current Web • Enable software services In short, “create a Web of Information”

4. Semantic Web Stack Source: W3C http://www.w3.org/2000/Talks/1206-xml2k-tbl/slide10-0.html

5. Debunking Semantic Web Myths Some Semantic Web myths … • Large Scale application of AI on the Web • Need to understand logic, knowledge representation techniques • Relies on centrally controlled ontologies • Meant to replace the current Web

6. Resource Description Framework • Data model of the Semantic Web • Central to the data integration requirements of Semantic Web • Allow data interchange without loss of meaning

7. RDF Basics • Data is the form of facts called statements • Each statement is a triple of resources of the form (Subject, Property, Object) e.g. (“Ayn Rand”, authorOf, “The Fountainhead”)

8. Resource There is a bit more to resources: • An entity in the data • Each resource identified by a URI (Universal Resource Identifier) • These URI ground the RDF to the Web • Provide the basis for comparison of resources, hence important • Resources can be literals too i.e. no URI associated with them

9. RDF as a graph • Triples • (index.html, dc:creator, staff-12345) • (index.html, dc:language, ”en”) • (index.html, creation-date, ”12 Aug 2006”)

10. How is RDF different from using XML? • Cleaner separation of syntax and semantics • Makes explicit the data model • However, XML is still used as a serialization mechanism • Other serialization mechanisms like Turtle and Notation 3 are available

11. Enter RDF-S (RDF-Schema) • RDF-S is to RDF what XML-Schema is to XML • RDF-S provides mechanisms to define domain vocabulary, just as a programming language allows you to define new types • Applications can then define vocabularies • Lightweight ontology

12. RDF-S Basics and Constructs RDF-S allows defining • resources as Classes, Properties or Datatypes • resources as instances of Classes • sub-class, sub-properties relations • domain/range of properties • resource containers like bag, list, etc.

13. A look at RDF-S • Animal rdfs:type rdfs:Class • Horse rdfs:subClassOf Animal • Dog rdfs:subClassOf Animal • number_of_legs rdfs:type rdfs:Property • number_of_legs rdfs:domain Animal • number_of_legs rdfs:range Integer

14. RDF in action • Dublin Core • vocabulary for describing networked, document like entities and their properties • RSS 1.0 • lightweight way of content syndication • SKOS (Simple Knowledge Organization System) • a model for expressing the basic structure and content of concept schemes. • XMP, FOAF, etc.

15. What is OWL? • Web Ontology Language, a W3C Recommendation • Derived from the DAML+OIL • Builds on RDF, uses RDF/XML • Knowledge Representation layer of the Semantic Web • An ontology: A formal conceptualization of the world • Concepts • Properties • Relationships • Additional Constraints

16. Why OWL? • Supply richer semantics and knowledge base for agents • Overcome limitations of RDF-S • Local Properties of classes • disjointness of classes • cardinality restrictions • ... and others

17. OWL Features – Class Level (1) • Class Equivalence • Faculty owl:equivalentClass AcademicStaffMember • Class Disjointness • Professor owl:disjointWith AssociateProfessor • Every class is a sub-class of owl:Thing and super-class of owl:Nothing

18. OWL Features: Class Level (2) • Boolean combinations to create new classes • Union • staff=owl:unionOf ( AcademicStaffMember, Non-AcademicStaffMember) • Intersection • TA=owl:intersectionOf (student , stipendee) • Complement • Non-AcademicStaffMember= owl:complementOf (AcademicStaffMember)

19. OWL Features: Property Level (3) • Inverse Relation • isTaughtBy owl:inverseOf teaches • Special Properties • Transitive • isBroaderThan owl:TransitiveProperty • Symmetric • isPartner owl:SymmetricProperty • also, Function, One-One

20. OWL Features: Property Level (4) • Property Restrictions: Constrain attributes to take only certain values for particular classes • Existential Quantification • academicStaffMember teaches underGraduateCourse • Universal Quantification • firstYearCourse isTaughtBy Professor • Cardinality Constraints • academicStaffMember takes a maximum of 3 courses

21. OWL Expressiveness • Powerful OWL constructs, but at a cost • Trade-off between • Expressiveness of language • Decidability and completeness • Tractability • 3 OWL flavours defined with different degrees of expressiveness

22. OWL Flavours • OWL Full • Most expressive, undecidable • Any RDF valid OWL Full • OWL DL • Based on Descriptive Logic • Guarantees computational completeness and decidability • OWL Lite • Good enough for simple taxonomies

23. Semantic Web & OWL in action • Knowledge Management • Web Services • Data Integration

24. Knowledge Management • Use Semantic Web technologies for • Better organization of information • Improved search • Portals: • Vodafone's Live Mobile Portal • Sun’s SwordFish • Organizing data in biosciences: GoPubMed

25. Web Services • Where does the Semantic Web come in? • Automatic Discovery of Web Services • Automatic Web Service Invocation • Automatic Web Service Composition and Interoperation • OWL-S markup language for describing services profiles, process models and grounding for interoperability