Scripting the Semantic Web

1
Scripting the Semantic Web

• Marian Babik, Ladislav Hluchy
• Intelligent and Knowledge-oriented Technologies
  Group
• Institute of Informatics, SAS

2
Goals of the presentation

• Semantic Web
• Object-oriented Systems
• Introduce deep integration ideas
• Show initial implementation of the Python
  integrated with OWL

3
Semantic Web

• Extension of the current Web, providing
  infrastructure for the integration of data on the
  Web
• Data should be available for further processing
• It should be possible to combine and merge data
  on a Web scale
• Metadata, reasoning
• Making data machine processable
• Names for resources
• Common data model RDF
• Access to the data SPARQL, OWL-QL
• Define vocabularies RDFS, OWL, SKOS
• Reasoning OWL, Rules

4
Description Logics, OWL-DL

• Class definition, axioms
• Complex class definitions based on the class
  descriptions
• Property definition, axioms
• RDF Schema constructs (rdfssubPropertyOf,
  rdfsdomain and rdfsrange)
• relations to other properties owlequivalentPrope
  rty and owlinverseOf)
• global cardinality constraints
  owlFunctionalProperty and owlInverseFunctionalPr
  operty
• logical property characteristics
  owlSymmetricProperty and owlTransitiveProperty
• Individuals
• facts about class membership and property values
  of individuals

5
Object-oriented systems

• Data Structure is an Object that contains both
  Data and Methods
• Relationships can exist between different Objects
  (inheritance etc)
• Modular and easily extendible frameworks
• Examples Java, C, Smalltalk, Python, Ruby, PHP
• Web Frameworks

6
Issues in mapping DL to OO

• Similarities between DL and OO
• Class Hierarchy w/ Multiple Inheritance
• DL Properties correspond to OO Class fields
• Differences between DL and OO
• Semantics
• Open Vs. Closed World Semantics
• Interpretations
• Implicit Vs. Explicit Set Membership
  Interpretation
• Extensions
• Automatic Inferences Vs. Manual Modifications
• Behavior
• None in DL, crucial to OO

7
Deep integration ideas

• Import OWL classes alongside classes defined
  normally (native API)
• Support for intensional definition of
  classes/properties (OWL statements)
• Separation of concerns among declarative and
  procedural aspects
• Similar to what SQLObject does for databases

8
Benefits

• Definition of data and domain using Semantic Web
  tools (Protégé, SWOOP)
• Native OWL API
• OWL Inference and Web sharing of concepts
• New programming paradigm for Python
• Native interface for ontologies, developers can
  work directly with OWL classes and their
  instances
• Existing Python Web frameworks and access to
  large set of libraries

Ontologies Python
9
Related Work

• Jena, OWL API
• Java is statically-typed language
• Different notion of polymorphism means APIs have
  to introduce sophisticated design patterns

Resource r myModel.getResource( myNS
"DigitalCamera" ) OntClass cls (OntClass)
r.as( OntClass.class ) Restriction rest
(Restriction) cls.as( Restriction.class )

• Jastor
• Automatic Mapping of OWL Ontologies into Java
  Kalyanpur, et.al
• Protégé Bean-Generator
• FIPA, Jade Poggi et.al.
• RDFReactor

10
Related Work

• Python
• RDFLib statement-centric (Daniel Krech)
• CWM model centric
• Pychinko model, statement-centric, rule based
• MetaLog statement, resource-centric, based on
  prolog
• Sparta, Tramp resource centric
• Others (see http//www.wiwiss.fu-berlin.de/suhl/b
  izer/toolkits/)
• Active RDF (Ruby) native mapping, RDF,
  integrated with Rails
• RDFHomepage (PHP) native mapping

11
OWL-DL and Python

• Ontology corresponds to Python module
• OWL Class corresponds to Python Class
• Instance of OWL Class corresponds to Python
  Object
• OWL Property corresponds to Python Class or
  Python method
• OWL DatatypeProperty corresponds to Attribute

12
Classes

• OWL Class corresponds to Python Class
• Instances of OWL Class correspond to Python
  Objects
• Class definition
• Each python class has two attributes
• URI (owlClass URI)
• definedBy (intensional definition)

gtgtgt from seth import Thing, Property gtgtgt Person
Thing.new( Person a owlClass .') gtgtgt print
Person ltclass seth.Meta.Persongt gtgtgt Man
Thing.new(' Man a owlClass rdfssubClassOf
Person .')
13
Classes (2)

• gtgtgt PersonWithSingleSon Thing.new("""PersonWithS
  ingleSon
• a owlClass
• rdfssubClassOf
• a owlRestriction owlcardinality "1"
• owlonProperty hasSon
• rdfssubClassOf
• a owlRestriction owlcardinality "1"
• owlonProperty hasChild .""")

gtgtgt model OntModel() gtgtgt model.bind(people,ht
tp//somewhere.net/people) gtgtgt
Thing.setOntModel(model) gtgtgt Person Thing.new(
peoplePerson a owlClass .)
14
Properties

• owlProperty corresponds to python objects
• Similar to Class definition
• But such classes dont have instances
• owlProperty corresponds to python object
  attribute (method)

gtgtgt hasChild Property.new('hasChild a
owlObjectProperty .') gtgtgt print hasChild ltclass
'seth.Meta.hasChild'gt gtgtgt hasSon
Property.new('hasSon a owlObjectProperty
rdfssubPropertyOf hasChild .')
gtgtgt John.hasChild(Bob)
15
Individuals

• Individual is python object (class instance)

gtgtgt Bob PersonWithSingleSon() gtgtgt John
Person() gtgtgt print John ltseth.Meta.Person object
at 0xb7a29e4cgt gtgtgt hasChild(Bob, John)
16
Ontologies

• owlOntology corresponds to Python module

gtgtgt ModuleFactory(people, http//somewhere.net/
people.owl) gtgtgt import people gtgtgt print
people.Person ltseth.people.Person Man a
owlClass rdfssubClassOf Person . gtgtgt print
people.John ltseth.people.Person object at
0xb7a29e4cgt

• serialization

gtgtgt people.serialize(/tmp/test.owl,
RDF/XML-ABBREV)
17
Queries

• Triple like queries

gtgtgt for individual in Person.findInstances() ...
print individual, individual.name ltseth.Meta.M
an object at 0xb7d0b64cgt Peter ltseth.Meta.Person
object at 0xb7d0b50cgt John ltseth.Meta.Person
object at 0xb7d0b6ecgt Jane gtgtgt for who in
hasSon.q(Bob) ... who.name 'John' gtgtgt print
hasSon.query(Bob, John) 1

• OWL-QL, SPARQL
• Native queries

18
Prototype architecture

• Interaction with Java-based libraries (reasoners)

19
Python and Java

• Jython - Python implemented for the Java Virtual
  Machine
• JPE(Java-Python Extension) - uses JNI to provide
  a bridging mechanism between Java and a Python
  interpreter
• JPype - interfacing at the native level in both
  Virtual Machines
• SPIRO - uses ORB (Object-Request Broker)
  technology
• GCJ

20
Issues

• Performance (conversion between JVMs)
• Cache
• Datatype reasoning
• Open world semantics
• Debugging
• Rules

21
Summary

• SETH Homepage
• http//seth-scripting.sourceforge.net
• Available through CVS under MIT license
• Discussion, support mailing list
• seth.ui_at_savba.sk
• Plans
• Contact
• Marian.Babik_at_gmail.com

22
References

• Introduction to Semantic Web
• http//www.w3.org/2006/Talks/0524-Edinburgh-IH/Ove
  rview.pdf
• Deep Integration
• http//www.semanticscripting.org/SFSW2005/papers/V
  randecic-Deep_Integration.pdf
• http//gigaton.thoughtworks.net/ofernand1/DeepInt
  egration.pdf
• http//www.semanticscripting.org/SFSW2006/Paper1.p
  df

23
Resource Description Framework (RDF)

• Based on triples (subject, predicate, object)
• Labelled connection btw. two resources (subject,
  object)
• Subject, predicate are URI-s Object is URI (e.g.
  Subject) or Literal
• ( http//www.marianbabik.org, http//.../myCalenda
  r, http//.../calendar )
• (subject, predicate, object) can be seen as a
  labeled edge in a graph
• Serialization XML, Turtle, N3, etc.

24
RDF example
ltrdfDescription rdfabout"http//www.tuke.sk
/fei-cit/babik"gt ltshasNamegtMarian
Babiklt/shasNamegt ltshasWritten
rdfresource"http//www.semanticscripting.org/SFS
W2006/Paper1.pdf"/gt lt/rdfDescriptiongt
Turtle (N3) XML/RDF

• lthttp//www.tuke.sk/fei-cit/babikgt
• shasName Marian Babik
• shasWritten lthttp//www.semanticscripting.org/
  SFSW2006/Paper1.pdfgt .

25
RDF Vocabulary Description Language (RDFS)

• RDF lacks vocabulary with semantic meaning
• RDFS defines resources and classes and
  relationships among classes/resource
• rdftype, rdfsClass, rdfssubClassOf
• rdfssubPropertyOf,
• Resource may belong to several classes (range,
  domain)
• Special kind of type system

26
RDFS example
Person rdfssubClassOf Man . John rdftype
Man .
ltrdfDescription rdfIDJohngt ltrdftype
rdfresourceMan/gt lt/rdfDescriptiongt ltrdfDescr
iption rdfIDMangt ltrdfssubClassOf
rdfresourcePerson/gt lt/rdfDescriptiongt
27
RDFS example (2)

• Person rdfssubClassOf Man .
• John rdftype Man .
• isProgrammer
• rdftype rdfProperty
• rdfdomain Person
• rdfrange rdfsLiteral .

28
Web Ontology Language (OWL)

• RDFS doesnt solve all the issues
• Reasoning about terms
• Behavior of properties (symmetric, transitive,
  etc.)
• Cardinality constraints
• Equivalence of classes
• Enumeration (oneOf), union of classes
• Datatype properties
• Compromise btw
• Rich semantics
• Feasibility, implementability
• Three layers
• OWL-Lite,
• OWL-DL (based on Description Logic SHOIN(D))
• OWL-Full (might be undecidable)

29
OWL-DL example
_at_prefix people lthttp//sample.org/peoplegt
. peoplePerson a owlClass . peopleMan a
owlClass rdfssubClassOf peoplePerson peopleha
sChild a owlProperty . peoplehasSon a
owlPropery rdfssubPropertyOf
peoplehasChild PersonWithSingleSon a owlClass
rdfssubClassOf a owlRestriction
owlcardinality "1" owlonProperty
hasSon rdfssubClassOf a
owlRestriction owlcardinality "1"
owlonProperty hasChild . PeopleJohn a
peoplePerson . PeopleBob a peoplePersonWithSing
leSon .