Semantics in the Semantic Web

1
Semantics in the Semantic Web

• Olivier Buchwalder - EPFL - LTI
• Doctoral Course
• Conceptual Modeling

2
Outline

• Motivation
• My Interests
• The Traditional Web
• The Semantic Web Approach
• The Semantic Web
• Technologies Overview
• Resource Description Framework (RDF)
• RDF Schema (RDFS)
• Web Ontology Language (OWL)
• Conclusion

3
Motivation My Interests

• Domain-Specific Modeling Environment
• Modeling and Implementing Web-Applications
  WebLang DSL
• Generic Facilities for creating new DSL
• Based on Textual Syntax for Edition
• Graphical Output for Documentation
• Connection
• Definition of domain-specific concepts
• Ontology is a central aspect of the Semantic Web

4
The Traditional Web

• Information available in HTML, XHTML
• natural language (English, French, Chinese, ...)
• images, videos, text, menu, table
• Human-Oriented
• decuce from context the relevant information
• mental associations between information
• mainly designed for human sensors
• Combining Information Meaning
• human can combine relevant information easily
  (time!)
• hotel, flight time table, weather available from
  different sites
• Search Engines too many false hits
• But the machines cant!

5
The Semantic Web Approach

• Moving from a Web of Document to a Web of Data
• Provides a common framework that allows data to
  be shared and reused across application,
  community.
• Allow automated agents to perform tasks for users
• Led by W3C with researchers and industrial
  partners
• Possible Applications
• Your automatic airline reservation
• knows about your preferences (user profile), your
  past
• can combine the local knowledge with remote
  services airline preferences, weather report,
  calendaring
• Database Integration
• very different in structure, in content
• most of these data are now accessible on the Web
  (private)

6
Technologies Overview

• Resources identifier URI
• Common data model to access, connect, describe
  RDF
• Data query SPARQL
• Common vocabularies (Ontologies) RDFS, OWL
• Reasoning logics OWL, SWRL

• W3C Recommendations in 2004 for following
  foundation specifications
• RDF, RDFS, OWL
• Standardization work is underway in
• Query (SPARQL), Rules (SWRL, ...)

7
RDF Resource Description Framework

• The Element of SW RDF Triples (s,p,v)
• labelled connection between two resources (s, p
  are URI-s and v is a URI or a literal)
• e.g. The sky has the color blue
• RDF is a general model for defining triples with
  machine readable formats (RDF/XML, Turtle, N3,
  ...)
• like hyperlinks on the traditional web, except
  that
• a relationship must have a name
• no notion of current document
• no attached user-interface action like for a
  hyperlink

8
SW RDF

• Set of RDF triples is a directed labeled graph
• nodes subjects, values (objects), edges
  properties

Everything
Tables
Trees
9
RDF/XML Example

• ltrdfDescription rdfabout "http//www.ivan-herm
  an.net" gt
• ltfirstnamegtIvanltfirstnamegt
• ltmyCalendar rdfresource "http//.../myCalenda
  r" /gt
• ltsurnamegtHermanltsurnamegt
• lt/rdfDescriptiongt

10
RDF/Turtle Example

• lthttp//www.ivan-herman.netgt
• firstName "Ivan"
• myCalendar lthttp//.../myCalendargt
• surname "Herman".

11
RDF Data Integration

• Easy to Merge Graphs
• Same URI-s refers to the same resource
• Node with identical URI are considered as
  identical
• Ontologies / rules for helping integration
• The RDF terms firstName, myCalendar
• Are they known?, correct?, the defined
  relationship are valid
• Need to formalize the adressed concepts, terms
  and relationships

12
RDF Schema (RDFS)

• Vocabulary Description Language (knowledge)
• Provides basic elements for the description of
  ontologies
• Class defining a resource as a class for other
  resources
• schema gt Dog rdftype rdfsClass
• data gt Fido rdftype Dog
• Subclass defining hierarchy of classes (many
  superclasses allowed)
• Dog rdfssubClassOf Animal
• Fido rdftype Animal (inferred from rules)
• Property constraint by domain range
• name rdftype rdfProperty
• Fido name "Fido"
• name rdfsdomain Dog
• name rdfsrange rdfsLiteral

13
RDFS Graphs
rdftype
Dog
Animal
rdfsClass
rdftype
rdftype
rdfsResource
rdfssubClassOf
rdftype
Fido
name
 Fido 
14
RDF/Turtle Example 2

• lthttp//www.ivan-herman.netgt
• foaffirstName "Ivan"
• abcmyCalendar lthttp//.../myCalendargt
• foafsurname "Herman".

15
RDFS Merging Graphs

• Extra RDFS statements provide the necessary glue
  for merging graphs
• aauthor same as fauteur (object node)
• both identify a Person (FOAF) gt use subclass
• identify aname (from node aauthor) with
  foafname gt use subproperty
• Merging capabilities
• Depends on the level of additional knowledge
  (glue) added to the data
• more complex ontologies gt more complex reasoning
  procedures

16
OWL Web Ontology Language

• More formal concepts and relationships
• Constraints on properties, possiblity to reason,
  deduce and check (Description Logic)
• if A is left of B and B is left of C, is A left
  of C?
• A layer on top of RDFS (DAML OIL gt OWL)
• Introduces its own Class and Thing to
  differentiate the classes from individuals
  (Confuse in RDFS)
• Construct classes from existing ones
• enumerate its content
• intersection, union, complement
• property restrictions
• disjointness or equivalence of classes

17
OWL Layers

• OWL Lite
• Classification hierarchy and simple constraints
  and logic.
• OWL DL (Description logic)
• Maximum expressiveness, computational
  completeness and decidability
• OWL Full
• Maximum expressiveness, syntactic freedom of RDF
  with no computational guarantees
• Differences with a database system
• Properties of an individual not only the class
  properties
• No assumption that everything is known
• Classes and properties can have multiple
  definitions (in different models)

18
OWL Example

• ltrdfProperty rdfID"name"gt
• ltrdfrangegt
• ltowlClassgt
• ltowloneOf rdfparseType"Collection"gt
• ltowlThing rdfID"Flipper"/gt
• ltowlThing rdfID"Joe"/gt
• ltowlThing rdfID"Mary"/gt
• lt/owloneOfgt
• lt/owlClassgt
• lt/rdfrangegt
• lt/rdfPropertygt

19
OWL Example 2

• ltowlClass rdfID"MarineMammal"gt
• ltowlunionOf rdfparseType"Collection"gt
  
• ltowlClass rdfabout"Dolphin"/gt
• ltowlClass rdfabout"Orca"/gt
• ltowlClass rdfabout"Whale"/gt
• lt/owlunionOfgt
• lt/owlClassgt

20
OWL Example 3
ltowlClass rdfID"Dolphin"gt ltrdfssubClassOf
rdfresource"Mammal"/gt
ltrdfssubClassOfgt ltowlRestrictiongt
ltowlonProperty rdfresource"livin
gIn"/gt ltowlallValuesFrom
rdfresource"UnionOfSeaAndAmazonas"gt
lt/owlRestrictiongt lt/rdfssubClassOfgt lt/owlCla
ssgt
21
Ontologies / Vocabularies

• Ontologies are usually developed and shared by
  communities
• FOAF designed to describe people, their
  interests and interconnections
• Dublin Core information resources, digital
  libraries, with extensions for rights,
  permissions
• MusicBrainz description of CDs, music tracks
• SKOS about knowledge systems
• SIOC Semantically-Interlinked Online Communities
• DOAP descriptions of software projects
• ...

22
Conclusion

• Semantic Web is in Progress
• Under Construction, big challenge
• The foundations are stable
• RDF / RDFS / OWL
• The hard work is to create the ontologies
• Sharing ontologies may be vital in the process
• For common users, just use existing ontologies
• SW Applications (currently centralized)
• RDF version of Wikipedia, MusicBrainz service
• SW Tools
• RDF IsaViz, RDFAuthor, Longwell
• OWL Protege 2000, SWOOP
• Trust and Proof

23

• Thank you for your attention...
• Any Questions ?
• Sources
• The Semantic Web, Scientific American, May 2001,
  Tim Berners-Lee, James Hendler and Ora Lassila
• Introduction to the Semantic Web, 2006-05-24,
  Ivan Herman, WWW2006 tutorial
• Semantic Web Building on What Exists, Tim
  Berners-Lee, MIT Information Technology Conference