WP2: Tools

1
WP2 Tools

• Raphael Volz
• Universität Fridericiana zu Karlsruhe (TH)

2
WonderWeb Tools
Tools now use OWL, the W3C standard for Web
ontologies
3
WonderWeb Tools
KAON SERVER Infrastructural KernelApplication
Server using Semantic technologies formanagement
of hosted components
4
KAON SERVER Functionality

• Functionality of common Application Server
  
• Flexible handling of Semantic Web
  modules(dynamic registering and unregistering)
• FaCT
• Racer
• Ontobroker
• KAON RDF stores
• KAON Ontology stores
• Sesame
• Extensible for future developments
• Generalization of RDF APIs (towards RDF Net API)

5
KAON SERVER Architecture
6
KAON SERVER Semantic technologies

• Ontology is used for
• component discovery
• API discovery
• classification of tools
• implementation tasks
• ? ASSW is semantically enhanced application
  server!

7
WonderWeb Tools
Lift Extract light-weight ontologies from legacy
information sourcesDB Schema, XML Schema, UML
Models, Java Documentation
8
OntoLiFT

• Describes heuristics for extracting light-weight
  ontologies from important legacy resources
• Relational Database Schema
• Logical Schema found in running DBs
• XML Schema Languages
• Generic formal approach building on Regular Tree
  Grammars
• UML Software Specifications
• Static Aspects of Software modelled in Class
  Diagrams
• JavaDoc Software Documentation
• Text Mining techniques to extract ontologies from
  textualDocumentation

9
LiFT Rel. DB Example
10
WonderWeb Tools
OilEd Visual Editor for creating OWL ontologies
11
What is OilEd?

• A simple ontology editor.
• Low cost, easy editor.
• A platform to explore how to use a reasoner.
• Originally intended as ademonstrator, now
  widelyused (gt2000 downloads).
• Initially targeted at OIL,supports DAMLOIL,and
  now supports OWL.

12
Reasoning

• KB translated to equivalent DL model, and passed
  to a reasoner.
• Spots inconsistent definitions
• e.g. contradictions in cardinality constraints or
  value restrictions.
• mad cows!
• Organises the classification hierarchy
• Discovering new superclasses.
• Particularly useful when using defined classes.
• Subtle side-effects
• Superclasses inferred due to domain/range
  restrictions.
• One shot connection to the reasoner.
• Allows temporary inconsistency
• Communication with reasoner via KAON SERVER

13
WonderWeb Tools
ReasonersMultiple systems (FACT,Hoolet, DLP)
to reason with ontologies Common Interface (DIG)
proposed and agreed upon (Racer, OWLP)
14
Reasoners

• Reasoners developed during the project (WP2) were
  crucial to the success of the standardisation
  activity.
• Why ? W3C standardisation requires demonstration
  of implementation experience, in particular
• FaCT
• Hoolet (1st Order reasoner)
• KAON DLP Prototype

15
DIG Interface

• A common interface to DL systems
• Supported by FACT, FACT, Racer, OWLP
• A new KRSS.
• XML Schema for concept language.
• Using HTTP for communication.
• Simple, Level 0 approach
• Sufficient to support simple, one-shot
  applications, e.g. OilEd
• Largely developed in WonderWeb project

16
WonderWeb Tools
OntoDiffOntology Change DetectionMethodology
plus practical tool support
17
Versioning tool OntoView

• Goal tool for (ordinary) ontology engineers
• helps to specify complete information about
  changes
• Functions
• compare versions of ontologies
• show propagation of change
• allows specification of conceptual relation
• distils the transformation operations
• patters of operations can be specified

18
Comparing ontology versions

• What should be compared?
• not textual representation (like diff)
• formatting is not important
• could be different representations for same
  construct
• not logical theory
• small change might affect whole ontology
• instead look at intended definition of classes
  and properties
• groups of axioms that form a definition
• follows ideas of ontology engineer
• gives a heuristic representation of the change

19
(No Transcript)
20
WonderWeb Tools
Further Tools Driving Theory and Practice of Web
Ontology Field
21
Thank you !
22
Change detection procedure

• Assume RDF-based ontology languages
• Ontologies are parsed and stored as RDF triples
• group statements resulting from one definition
  (first level XML tree)
• identify group by rdfID / rdfabout statement

ltowlClass rdfID"Person"gt ltrdfssubClassOf
rdfresource"Animal"/gt ltrdfssubClassOfgt
ltowlRestrictiongt ltowlonProperty
rdfresource"hasParent"/gt
ltowlallValuesFrom rdfresource"Person"/gt
lt/owlRestrictiongt lt/rdfssubClassOfgt lt/owlClas
sgt
about Person
Person rdftype damlClass Person rdfssu
bClassOf Animal Person rdfssubClassOf anon_1
anon_1 rdftype damlRestriction anon_1 owlonPr
operty hasParent anon_1 owlallValuesFrom Person
23
Change detection - 2

• Versions are compared by aligning groups of
  statements about same concept / property
• order is used in case of multiple occurrences

v1
v2
about Person
about Person
Person rdftype owlClass Person rdfssubClassO
f Animal Person rdfssubClassOf anon_1
anon_1 rdftype owlRestriction anon_1 owlonPro
perty hasParent anon_1 owltoClass Person
Person rdftype owlClass Person rdfssubClassO
f Animal
about parentOf
about parentOf
parentOf rdftype owlProperty parentOf rdftype
owlObjectProperty
parentOf rdftype owlProperty
parentOf rdftype owlObjectProperty
24
Change detection - 3
IF existold ltA, Y, Zgt existnew
ltX, Y, Zgt not-existnew ltX, Y, Zgt THEN
change-type A

• Rules are used to describe change types
• language specific
• requires computation of RDFS closure
• i.e. all statements should be made explicit
• Example find a change in a OWL slot restriction

IF existold ltX, rdfssubClassOf, Y1gt
ltY1, rdftype, owlRestrictiongt ltY1,
owlonProperty, Y2gt ltY1, owlallValuesFrom,
Zgt existnew ltX, rdfssubClassOf, Y1gt
ltY1, rdftype, owlRestrictiongt ltY1,
owlonProperty, Y2gt not-existnew ltY1,
damltoClass, Zgt THEN logicalChange.localPropertyV
alue X
25
Change detection - 4

• Original file is used for visualisation of change
• user specific formatting is preserved
• Strong points
• generic mechanism
• can be used with all RDFS based languages
• language specific parts are encoded in rules
• RDF data model hides a lot of syntactic changes
• existing tools are used for parsing, state
  maintenance and computation of RDFS closure