Ontologies and the Semantic Webbeaune/websem/ontoSemWebStaab.pdf · Origin and History • Ontology in Philosophy • a philosophical discipline, branch of philosophy ... Introduction

Steffen Staab (1)ISWeb – Informationssysteme & Semantic Web

Ontologies and the Semantic Web

SMBM-2006Jena, April 9, 2006

Steffen Staabhttp://isweb.uni-koblenz.de


Agenda

1. Ontologies2. Semantic Web3. Semantic Web Languages4. Some Applications (Ontoprise)5. Ontologies & Text


Part I

Introduction to Ontologies


Origin and History

• Ontology in Philosophy• a philosophical discipline, branch of philosophy

that deals with the nature and the organization of reality

• Science of Being (Aristotle, Metaphysics, IV, 1)

• Tries to answer the questions:• What characterizes being?• Eventually, what is being?


Aristotle - Ontology• Before: study of the nature of being

• Since Aristotle: study of knowledge representation and reasoning

• Terminology:– Genus: (Classes)– Species: (Subclasses)– Differentiae: (Characteristics which allow to group or

distinguish objects from each other)• Syllogisms (Inference Rules)


Example for differentiae (adapted from Uta Priss, in preparation)

XXXOsmond

XXXCopito

XXNemo

XXXBugs Bunny

XXXSnoopy

XXXGarfield

mammalkoalagorillafishrabbitdogcatcartoonreal


Organizing the Objects as a Lattice


What is an Ontology?Gruber 93:

An Ontology is aformal specificationof a sharedconceptualizationof a domain of interest

⇒ Executable, Discussable⇒ Group of persons⇒ About concepts⇒ Between application

and „unique truth“


Why Develop an Ontology?• To make domain assumptions explicit

– Easier to change domain assumptions– Easier to understand and update legacy data

• To separate domain knowledge from operational knowledge– Re-use domain and operational knowledge

separately

• A community reference for applications

• To share a consistent understanding of what information means


Taxonomy

Object

Person Topic Document

ResearcherStudent Semantics

OntologyDoctoral Student

Taxonomy := Segmentation, classification and ordering of elements into a classification system according to theirrelationships between each other

PhD Student F-Logic

Menu


Thesaurus

Object



PhD StudentDoktoral Student

• Terminology for specific domain• Taxonomy plus fixed relationships (similar, synonym, related to) • originate from bibliography

similarsynonym

OntologyF-Logic

Menu


Topic Map

Object



PhD StudentDoktoral Student

knows described_in

writes

AffiliationTel

• Topics (nodes), relationships and occurences (to documents)• ISO-Standard• typically for navigation- and visualisation

OntologyF-Logic

similarsynonym

Menu


OntologyF-Logic

similar

OntologyF-Logic

similarPhD StudentDoktoral Student

Ontology (in our sense)

Object


Tel

PhD StudentPhD Student

Semantics

knows described_in

writes

Affiliationdescribed_in is_about

knowsP writes D is_about T P T

DT T D

Rules

subTopicOf

• Representation Language: Predicate Logic (F-Logic)• Standards: RDF(S); OWL

ResearcherStudent

instance_of-1

is_a-1

is_a-1

is_a-1

Affiliation

Affiliation

York Sure

AIFB+49 721 608 6592


Ontologies - Some Examples• General purpose ontologies:

– DOLCE, http://www.loa-cnr.it/DOLCE.html– The Upper Cyc Ontology, http://www.cyc.com/cyc-2-1/index.html– IEEE Standard Upper Ontology, http://suo.ieee.org/

• Domain and application-specific ontologies:– GALEN, http://www.openclinical.org/prj_galen.html– Foundational Model of Anatomy, http://sig.biostr.washington.edu/projects/fm/AboutFM.html– RETSINA Calendering Agent, http://ilrt.org/discovery/2001/06/schemas/ical-full/hybrid.rdf– Dublin Core, http://dublincore.org/

• Semantic Desktop Ontologies– Semantics-Aware instant Messaging: SAM Ontology,

http://www.uni-koblenz.de/FB4/Institutes/IFI/AGStaab/Research/sam– Haystack, http://haystack.lcs.mit.edu/– Gnowsis, http://www.gnowsis.org/– Piggybank, http://simile.mit.edu/piggy-bank/

• Web Services Ontologies– Core ontology of services http://cos.ontoware.org– Web Service Modeling ontology http://www.wsmo.org– OWL-S, http://www.daml.org/services/owl-s/1.0/

• Ontologies in a wider sense– GO - Gene Ontology, http://www.geneontology.org/– UMLS, http://www.nlm.nih.gov/research/umls/– Agrovoc, http://www.fao.org/agrovoc/– Art and Architecture, http://www.getty.edu/research/tools/vocabulary/aat/– DTD standardizations, e.g. HR-XML, http://www.hr-xml.org/– WordNet / EuroWordNet, http://www.cogsci.princeton.edu/~wn


Ontologies and Their Relatives

Catalog / ID

Terms/Glossary

Thesauri

InformalIs-a

FormalIs-a

FormalInstance

Frames

ValueRestric-tions

Generallogical

constraints

AxiomsDisjointInverseRelations,...


Ontologies and Their Relatives (cont´d)Front-End

Back-End

Topic Maps

Extended ER-Models

Thesauri

Predicate Logic

Semantic Networks

Taxonomies

Ontologies

Navigation

Queries

Sharing of Knowledge

Information Retrieval

Query Expansion

Mediation Reasoning

Consistency CheckingEAI


Applications of Ontologies• Natural Language Processing and Machine Translation, e.g. Nirenburg et al.

2004, Maedche et al. 2001, Agirre et al. 1996, Beale et al. 1995• Semantic Web, see http://www.w3.org/2001/sw/ and

http://www.w3.org/2001/sw/WebOnt/• Knowledge Engineering & Management, e.g. Fensel 2001, Mullholland et al.

2000; Staab & Schnurr, 2000; Sure et al., 2000, Abecker et al. 1997• Electronic Commerce, e.g. RosettaNet3 and Ontology.org4• Information Retrieval and Information Integration, e.g. Kashyap, 1999; Mena

et al., 1998; Wiederhold, 1992• Intelligent Search Engines, e.g. WebKB (Martin et al. 2000), SHOE (Heflin &

Hendler, 2000), OntoSeek (Guarino et al., 1999), Ontobroker (Decker et al., 1999)

• Digital Libraries, e.g. Amann & Fundulaki, 1999• Enhanced User Interfaces, e.g. (Kesseler, 1996), Inxight5• Software Agents, e.g. OnTo-agents, FIPA, (Gluschko et al., 1999; Smith &

Poulter, 1999)• Business Process Modeling, e.g. Decker et al., 1997; TOVE, 1995; Uschold et

al., 1998


Overview Literature

S. Staab, R. Studer. Handbook on Ontologies. Springer, 2004.


Part II

Introduction to Semantic Web


Syntax is not sufficient

Andreas

• Tel

• E-Mail


Information Convergence• Convergence not just in devices, also in “information”

– Your personal information (phone, PDA,…)Calendar, photo, home page, files…

– Your “professional” life (laptop, desktop, … Grid)Web site, publications, files, databases, …

– Your “community” contexts (Web)Hobbies, blogs, fanfic, social networks…

• The Web teaches us that people will work to share– How do we CREATE, SEARCH, and BROWSE in

the non-text based parts of our lives?


WWW vs. Semantic Web WWW :=

Hypertext &

Internet &

Social Phenomenon

Semantic Web :=

Semantic Web Language/Data &

Ontologies &

Internet &

Social Phenomenon


XML is unspecific:No predetermined vocabularyNo semantics for relationships

& must be specified upfront

Only possible in close cooperations– Small, reasonably stable group– Common interests or authorities

Not possible in the Web or on a broad scale in general !

Let’s try XML


CV

name

education

work

private

Meaning of Informationen:(or: what it means to be a computer)


CV

name

education

work

private

< >

< >

< >

< >

< >

< Χς >

< ναµε >

<εδυχατιον>

<ωορκ>

<πριϖατε>

XML ≠ Meaning, XML = Structure


Some Principal Ideas

• URI – uniform resource identifiers• XML – common syntax• Interlinked• Layers of semantics –

from database to knowledge base to proofs

Design principles of WWW applied to Semantics!!

Tim Berners-Lee, Weaving

the Web


The Semantic Web on one Slide

EmployeeEmployee

PostDocPostDoc ProfessorProfessor

PersonPerson

rdfs:subClass rdfs:subClass

rdfs:subClass

cooperatesWithcooperatesWith

rdfs:Rangerdfs:DomainOntology

<swrc:Professorrdf:ID="person_sst">

<swrc:name>Steffen Staab</swrc:name>

...</swrc:Professor>

http://www.uni-koblenz.de/~staab

rdf:typerdf:type

Meta-data

<swrc:PostDoc rdf:ID="person_sha"><swrc:name>Siegfried Handschuh</swrc:name>

...</swrc:PostDoc>

Webpage

http://www.deri.ie/~shaURL

<swrc:cooperatesWith rdf:resource = "http://www.uni-koblenz.de/~staab/

#person_sst"/>

swrc:cooperatesWith


The Semantic Web - Inference

swrc:name

swrc:member

swrc:homepage

swrc:cooperatesWith

swrc:member

swrc:project

swrc:project

DERI

Handschuh

swrc:affiliation

Visualization of a Logic Representation:

OWL, F-Logic, etc.

Nepomuk


Proof

The new Semantic Web Stack

URI Unicode

RDF Core

Spa

rQL

RDF Schema

DLP bit of OWL/Rule Enc

rypt

ion

Sig

natu

re

OWL Rules

Trust

Logic framework

XML Namespaces

Tim Berners-Lee, ISWC November 2005, http://www.w3.org/2005/Talks/1110-iswc-tbl/#(12)


Knowledge Provisioning


Tools for markup...

PhotoStuff Demo


Semi-automatic


Not tied to specific domains

M-OntoMat is publicly available http://acemedia.org/aceMedia/results/software/m-ontomat-annotizer.html

Shape erasure

Shape Color selection

Visual Descriptor selection

Draw panel

Descriptor extraction

Shape selection

VDE plug-in launch

Domain Ontology Browser Selected

region

Save Prototype Instances


Shared Workspace(Xarop + Screenshot)


Social networks:e.g. Friend of a Friend (FOAF)

• Say stuff about yourself (or others) in OWL files, link to who you “know”

Estimates of the number of Foaf users range from 2M-5M


Using FOAF in other contexts

http://trust.mindswap.orgJennifer Golbeck


Get a B&N price (In Euros)


Of a particular book


In its German edition?



Now.• RDF, RDFS and OWL are ready for prime time

– Designs are stable, implementations maturing

• Major Research investment translating into application development and commercial spinoffs

– Adobe 6.0 embraces RDF

– IBM releases tools, data and partnering

– HP extending Jena to OWL

– OWL Engines by Ontoprise GmbH, Network Inference, Racer GmbH

– Ontoprise is a strategic partner for Oracle and Software AG

– Proprietary OWL ontologies for vertical markets• c.f. pharmacology, HMO/health care, ... Soft drinks


Now: Plenty of annotations –unfortunately, not in the open

• Taggings are daily practice:– Flickr, http://www.flickr.com/– Delicious, http://del.icio.us/– Cite-u-like, http://www.citeulike.org/– Bibsonomy,…

• Plenty of annotations– Dooyoo, E-pinions– Quipe, http://www.quipe.com/ – Froogle, http://froogle.google.com/– Google Base, http://base.google.com/– RSS– E-Science data curation,

http://www.jisc.ac.uk/index.cfm?name=pub_escience– Semantic Wikis

• Web 2.0– would be easier with Semantic Web!


The Semantic Wave

(Berners-Lee, 03)

YOUARE

HERE2006

YOUARE

HERE2003


Semantic Technologies vs. Semantic Web

Semantic Technologies• Used by „Early Adopters“

• Mature

– Deductive Databases(Research since early 80ies)

– Description logics(Research since late 70ies)

- Ontobroker (Research prototypesince 1990; commercial since 1999)

• A lot of knowledge about integration withexisting technology (databases, modelling, …)

Semantic Web• Still „research-oriented“

• Currently: Used in Intranets

• Currently: Used for internetapplications with simple ontologies (Dublin Core, RSS, PICS, FOAF,…)

• Quite some way to go for fullfledged success, initial take-upnow by some focus groups


Application areas for SemanticTechnologies

• Software engineering: conceptual approachesneed semantic interchange language

• Data description:– Databases in bioinformatics– Multimedia data (complementary to MPEG 7/21)

• Data integration: data exchange benefits fromsemantic interchange language

• „Plug n‘play“ for dynamic(not necessarily „automatic“!!!) business process configuration: needs rich semantic descriptions


Prospectives of Semantic Web or

WWW vs. Semantic Web revisited

WWW :=

Hypertext &

Internet &

Social Phenomenon

Semantic Web :=

Semantic Web Language/Data &

Ontologies &

Internet &

Social Phenomenon

WithoutSocial Phenomenon

= Intranet

WithoutSocial Phenomenon

= Semantic DataIntegration

New and important

paradigms at their time, but

„less“ outreach


„Less“ vs „More“ Outreach

„Less“ equals a multi-billion dollar market

„More“ equals a change as radical as triggered by the WWW


Overview Literature

Frank van Harmelen, Grigoris Antinou. Semantic Web Primer, MIT Press 2005.


Part III

Semantic Web Languages


RDF


RDF Data Model• Resources

– A resource is a thing you talk about (can reference)– Resources have URI’s– RDF definitions are itself Resources (linkage)

• Properties – slots, defines relationship to other resources or

atomic values• Statements

– “Resource has Property with Value”– (Values can be resources or atomic XML data)

• Similar to Frame Systems


A simple Example

• Statement– “Ora Lassila is the creator of the resource

http://www.w3.org/Home/Lassila”

• Structure– Resource (subject)

http://www.w3.org/Home/Lassila– Property (predicate) http://www.schema.org/#Creator– Value (object) "Ora Lassila”

• Directed graphhttp://www.w3.org/Home/Lassila s:Creator Ora Lassila


Another Example

• To add properties to Creator, point through a intermediate Resource.

http://www.w3.org/Home/Lassila

s:Creator

Person://fi/654645635

Name

Ora Lassila [email protected]

Email


Collection Containers

• Multiple occurrences of the same PropertyTypedoesn’t establish a relation between the values– The Millers own a boat, a bike, and a TV set– The Millers need (a car or a truck)– (Sarah and Bob) bought a new car

• RDF defines three special Resources:– Bag unordered values rdf:Bag– Sequence ordered values rdf:Seq– Alternative single value rdf:Alt

• Core RDF does not enforce ‘set’ semantics amongst values


Example: BagThe students in

course 6.001are Amy, Tim,John, Mary,and Sue

Rdf:Bag

/Students/Amy

/Students/Tim

/Students/John

/Students/Mary

/Students/Sue

bagid1

/courses/6.001

students

rdf:type

rdf:_1

rdf:_2

rdf:_3

rdf:_4

rdf:_5


Example: Alternative• The source code for X11 may be found at

ftp.x.org, ftp.cs.purdue.edu, or ftp.eu.net

http://x.org/package/X11rdf:Alt

ftp.x.org

ftp.cs.purdue.edu

ftp.eu.net

altid

rdf:type

rdf:_1

rdf:_2

rdf:_3


Statements about Statements (Requirement 2: Dispute Statements)

• Making statements about statementsrequires a process for transforming them into Resources– subject the original referent– predicate the original property type– object the original value– type rdf:Statement


Example: Reification

• Ralph Swick believes that – the creator of the resource

http://www.w3.org/Home/Lassila is OraLassila

rdf:Statement

rdf:type

genid1

Ralph Swick

b:believedBy


rdf:subject

Ora Lassila

rdf:object

s:Creatorrdf:predicate

s:Creator


RDF Syntax I• Datamodel does not enforce particular syntax• Specification suggests many different syntaxes

based on XML• General form:

<rdf:RDF><rdf:Description about="http://www.w3.org/Home/Lassila">

<s:Creator>Ora Lassila</s:Creator><s:createdWith rdf:resource=“http://www.w3c.org/amaya”/>

</rdf:Description></rdf:RDF>

Starts an RDF-Description

Properties

Subject (OID)

Literal

Resource (possibly another RDF-description)


Resulting Graph

<rdf:RDF><rdf:Description about="http://www.w3.org/Home/Lassila">

<s:Creator>Ora Lassila</s:Creator><s:createdWith rdf:resource=“http://www.w3c.org/amaya”/>

</rdf:Description></rdf:RDF>

http://www.w3c.org/amaya


Ora Lassila

s:createdWiths:Creator


RDF Syntax II: Syntactic Varieties

<s:Homepage rdf:about="http://www.w3.org/Home/Lassila”s:Creator=“Ora Lassila”/>

<s:Title>Ora's Home Page</s:Title><s:createdWith><s:HTMLEditor rdf:about=“http://www.w3c.org/amaya”/>

</s:createdWith> </s:Homepage>

Typing Information In-Element Property

Property

Subject (OID)

http://www.w3c.org/amaya


Ora Lassila

s:createdWiths:Creator

HTMLEditor

s:Homepagerdf:type

rdf:type


RDF Schema (RDFS)

• RDF just defines the datamodel• Need for definition of vocabularies for the

datamodel - an Ontology Language!• RDF schemas are Web resources (and

have URIs) and can be described using RDF


RDF-Schema: Example

rdfs:Resource

xyz:MotorVehicle rdfs:Class

s s t

t

xyz:Truck

s

t

xyz:PassengerVehicle

s = rdfs:subClassOf t = rdf:type

xyz:Van s

s

xyz:MiniVan s

s

t t

t


Rdfs:subclassOf<rdfs:description about=„Xyz:Minivan“>

<rdfs:subclassOf about=„xyz:Van“/></rdfs:description><rdfs:description about=„myvan“>

<rdf:type about=„xyz:MiniVan“/></rdfs:description>

Predicate Logic Consequences:

Forall X: type(X,MiniVan) -> type(X,Van). Forall X: subclassOf(X,MiniVan) -> subclassOf(X,Van).


Rdf:property<rdf:description about=„possesses“>

<rdf:type about=„....property“/><rdfs:domain about=„person“/><rdfs:range about=„vehicle“/>

</rdf:description><rdf:description about=„peter“>

<possesses>petersminivan</possesses></rdf:description>

Predicate Logic Consequences:Forall X,Y: possesses (X,Y) -> (type(X,person) &

type(Y,vehicle)).


OWL


OWL

• OWL1.0 (acronym for Web OntologyLanguage) is a W3C recommendation

• OWL stems from a family of logics, called„description logics“


Description Logics(Terminological Logics, DLs)

• Fragments of FOL• Most often decidable• Moderately expressive• Stem from semantic networks• W3C Standard OWL DL corresponds to

SHOIN(D)


DLs – general structure• DLs are a Family of logic-based formalism for

knowledge representation • Special language characterized by:

– Constructors to define complex concepts and roles based on simpler ones.

– Set of axiom to express facts using concepts, roles and individuals.

• ALC is the smallest DL, which is propositionally closed: – ∧, ∨, ¬ are constructors, noted by u, t, ¬.– Quantors define how roles are to be interpreted:

Man u ∃hasChild.Female u ∃hasChild.Maleu ∀hasChild.(Rich t Happy)


Further DL concepts and role constructors

• Number restrictions (cardinality constraints) for roles:≥3 hasChild, ·1hasMother

• Qualified number restrictions:≥2 hasChild.Female, ·1 hasParent.Male

• Nominals (definition by extension): {Italy, France, Spain}

• Concrete domains (datatypes): hasAge.(≥21)

• Inverse roles: hasChild– ≡ hasParent• Transitive roles: hasAncestor* (descendant)• Role composition: hasParent.hasBrother (uncle)


DL Knowledge Bases• DL Knowledge Bases consist of two parts (in general):

– TBox: Axioms, describing the structure of a modelled domain (conceptual schema):

• HappyFather ≡ Man u ∃hasChild.Female u …• Elephant v Animal u Large u Grey• transitive(hasAncestor)

– Abox: Axiome describing concrete situations (data, facts):• HappyFather(John)• hasChild(John, Mary)

• The distinction between TBox/ABox does not have a deep logical distinction… but it is common useful modelling practice.


General DL Architecture

Knowledge Base

Tbox (schema)

Abox (data)

Man ≡ Human u Male

Happy-Father ≡ Man u ∃ has-child.Female u …

Happy-Father(John)

has-child(John, Mary) Infe

ren

ce S

yst

em

Inte

rface


Knowledge modelling in OWLExample ontology and conclusion from

http://owl.man.ac.uk/2003/why/latest/#2• Also an example for OWL Abstract Syntax.

Namespace(a = <http://cohse.semanticweb.org/ontologies/people#>)Ontology(

ObjectProperty(a:drives) ObjectProperty(a:eaten_by) ObjectProperty(a:eats inverseOf(a:eaten_by) domain(a:animal)) …Class(a:adult partial annotation(rdfs:comment "Things that are adult.") Class(a:animal partial restriction(a:eats someValuesFrom (owl:Thing))) Class(a:animal_lover complete intersectionOf(restriction(a:has_pet

minCardinality(3)) a:person)) …)


Knowledge modelling: examplesClass(a:bus_driver complete intersectionOf(a:person

restriction(a:drives someValuesFrom (a:bus))))

Class(a:driver complete intersectionOf(a:personrestriction(a:drives someValuesFrom (a:vehicle))))

Class(a:bus partial a:vehicle)

• A bus driver is a person that drives a bus. • A bus is a vehicle. • A bus driver drives a vehicle, so must be a driver. The subclass is inferred due to subclasses being used

in existential quantification.

bus_driver ≡ person u ∃drives.bus

bus v vehicle

driver ≡ person u ∃drives.vehicle


Knowledge modelling: examplesClass(a:driver complete intersectionOf(a:person restriction(a:drives

someValuesFrom (a:vehicle))))

Class(a:driver partial a:adult)

Class(a:grownup complete intersectionOf(a:adult a:person))

• Drivers are defined as persons that drive cars (complete definition) • We also know that drivers are adults (partial definition) • So all drivers must be adult persons (e.g. grownups)

An example of axioms being used to assert additional necessaryinformation about a class. We do not need to know that a driver isan adult in order to recognize one, but once we have recognized a driver, we know that they must be adult.

driver ≡ person u ∃drives.vehicle

driver v adult

grownup ≡ adult u person


Knowledge modelling: ExamplesClass(a:cow partial a:vegetarian)DisjointClasses(unionOf(restriction(a:part_of someValuesFrom

(a:animal)) a:animal) unionOf(a:plant restriction(a:part_ofsomeValuesFrom (a:plant))))

Class(a:vegetarian complete intersectionOf( restriction(a:eatsallValuesFrom (complementOf(restriction(a:part_ofsomeValuesFrom (a:animal))))) restriction(a:eats allValuesFrom(complementOf(a:animal))) a:animal))

Class(a:mad_cow complete intersectionOf(a:cow restriction(a:eatssomeValuesFrom (intersectionOf(restriction(a:part_ofsomeValuesFrom (a:sheep)) a:brain)))))

Class(a:sheep partial a:animal restriction(a:eats allValuesFrom(a:grass)))

• Cows are naturally vegetarians• A mad cow is one that has been eating sheeps brains• Sheep are animalsThus a mad cow has been eating part of an animal, which is

inconsistent with the definition of a vegetarian

∃partof.animal t animal ≡ plant t ∃partof.plant/


Knowledge modelling: ExampleIndividual(a:Walt type(a:person) value(a:has_pet a:Huey)

value(a:has_pet a:Louie) value(a:has_pet a:Dewey)) Individual(a:Huey type(a:duck)) Individual(a:Dewey type(a:duck)) Individual(a:Louie type(a:duck)) DifferentIndividuals(a:Huey a:Dewey a:Louie) Class(a:animal_lover complete intersectionOf(a:person

restriction(a:has_pet minCardinality(3))))ObjectProperty(a:has_pet domain(a:person) range(a:animal))

• Walt has pets Huey, Dewey and Louie. • Huey, Dewey and Louie are all distinct individuals. • Walt has at least three pets and is thus an animal lover.

Note that in this case, we don’t actually need to include person in the definition of animal lover (as the domain restriction will allowus to draw this inference).


Knowledge modelling: SomeResearch Challenges

• Concluding with– uncertainty (fuzzy, probabilistic)– Inkonsistencies (paraconsistent)– Rules– Further AI-Paradigms (nonmonotonic reasoning,

preferences …)• Maintenance (updates, infrastructure, etc)• Scalability of reasoning• …

www.ontoprise.de

© 2006 ontoprise GmbH - 1 -Home | Menu | Technology | References | End

Application Scenario: Semantic Inference

www.ontoprise.de


Background

Complex dependencies decrease the speed of development

Knowledge is distributed over different departments

Goal

Design of a Semantic Guide for capturing the dependenciesConfiguration of components

Integration into existing order systemEngineers can concentrate on creative

efforts

Audi: Semantic Testcar Configuration

www.ontoprise.de


Inference: to conclude implicit facts

Testengine has been tested and released

Chassis 17 is suited for 110 KW

Testengine has 104 KW

Rule 1: A Chassis has to be suited for the power of the engine

Rule 2: All parts have to be tested and released

Testengine is ready to test in a car

Fit of Testengine and Chassis 17

www.ontoprise.de


Application Scenario:Semantic Data Integration and Search

www.ontoprise.de


Knowledge Management for your Projects

Users keep their establishedsoftware tools

A knowledge model (ontology) both integrates and structuresthe information

The ontology is enriched withspecific expertise

The ontology empowers a context-aware and easy-to-usesearch and navigationsystem

All information to stay in theiroriginal place

? ??

Object


TechnicianDecisionMaker Content ApplicationMetho-

dology

www.ontoprise.de


GoalsImprove the effectivity and quality of work

Integrated serach over multiple sources

Usage of an ontology to improve results

Simple interface

Proof of Concept for SemanticWebtechnology for whole group

FactsUsers: 1000 peopleproject duration: 2 months

Deutsche Post IT Solutions (DHL group)

A Companywide Search- & Integration-Project

www.ontoprise.de


Editorial Process for Ontology Evolution

www.ontoprise.de


Application Scenario:Semantic Data Integration - II

www.ontoprise.de


Integration Problems

Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de



Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de



Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de



Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de



Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de



Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de


Vielfältige IntegrationsproblemeIntegration Problems

Value Conflict

Structure Conflict

Name Conflict

Languages

Duplicates

Missing Information

Multiple Interfaces

www.ontoprise.de


Import and Mapping of DB-Structures

www.ontoprise.de


Application Scenario:Intelligent Question Answering

www.ontoprise.de


Background

• Development of a Digital Aristotle• Phase 1 successfully closed in 2003• Phase 2 since January 2004

Functions

• Capturing of extensive set of chemical knowledge

• System passed the „Advanced Placement Test“

• Query is answered and answer is explained

Vulcan Inc: OntoBroker passes Advanced Placement Test

www.ontoprise.de


Ontobroker™ passed the Advanced Placement Test!Correct AnswersCorrect Explanations

PerformanceCYCORP 1650 MinutesStudent 240 MinutesStanford Research 38 MinutesOntoprise 9 Minutes


Semantic Web Applications –on the Internet


Conceptual architecture for semantic portal

Sources

Commondata model

CommonSemantics

Presentation& Use

Selection

P2PRelationalDatabase

...

K-EdutellaWrapper

Rel-DBWrapper

FileS

X(HT)ML Wrapper

Integration

OntologIE

Presen-tationView

Presen-tationView

View

HTML Page

RDF Output

HTML Form

...

...

QueryAPI

RDF API

RDF

InputView

Datenbank

Navi-Gation(HTML)

Navi-gationView

[IEEE Data Engineering, 2002]


OntoWeb Community

AnnotatedWeb PagesGenerated

Content Objects

Participating Site2

{ }

Participating Siten

{ }

Participating Site1

{ }

...

OntologyBrowse & Query

Front End

ContentSyndication

Service

http://www.ontoweb.org

OntoWeb-Portal[CRIS 2002]

EU IST Projekt OntoWeb


P2P Application: Bibster


Ontologies & Text

Part V


OL from Text as Reverse Engineering

Reverse Engineering

Write

Shared World Model


Some pre-History of Ontology Learning

• AI: Knowledge Acquisition– Since 60s/70s: Semantic Network Extraction and similar for Story Understanding

• Systems: e.g. MARGIE (Schank et al., 1973), LUNAR (Woods, 1973)

• NLP: Lexical Knowledge Extraction– 70s/80s: Extraction of Lexical Semantic Representations from Machine

Readable Dictionaries• Systems: e.g. ACQUILEX LKB (Copestake et al.)

– 80s/90s: Extraction of Semantic Lexicons from Corpora for Information Extraction Systems

• Systems: e.g. AutoSlog (Riloff, 1993), CRYSTAL (Soderland et al., 1995)

• IR: Thesaurus Extraction– Since 60s: Extraction of Keywords, Thesauri and Controlled Vocabularies

• Based on construction and use of thesauri in IR (Sparck-Jones, 1966/1986, 1971)• Systems: e.g. Sextant (Grefenstette, 1992), DR-Link (Liddy, 1994)


Some Current Work on Ontology Learning from TextTerm Extraction

• Statistical Analysis• Patterns• (Shallow) Linguistic Parsing• Term Disambiguation & Compositional Interpretation• Combinations

Taxonomy Extraction• Statistical Analysis & Clustering (e.g. FCA)• Patterns• (Shallow) Linguistic Parsing• WordNet• Combinations

Relation Extraction• Anonymous Relations (e.g. with Association Rules)• Named Relations (Linguistic Parsing)• (Linguistic) Compound Analysis• Web Mining, Social Network Analysis• Combinations

Relation Label Extraction• Extension of Association Rules Algorithm

Definition Extraction• (Linguistic) Compound Analysis (incl. WordNet)


Some Current Work on Ontology Learning from Text

AIFB – TextToOnto (Maedche and Staab, 2000; Cimiano et al., 2005)– Term Extraction and Taxonomy Extraction

• Statistical Analysis• Conceptual Clustering (FCA), Patterns, WordNet (+ Combination)

– Relation Extraction• Anonymous Relations (Association Rules)• Named Relations (Subcategorization Frames)

CNTS Univ. Antwerpen, VUB (Reinberger et al., 2004)– Concept Formation + Relation Extraction

• Shallow Linguistic Parsing• Clustering

DFKI – OntoLT (Buitelaar et al., 2004), RelExt (Schutz and Buitelaar, 2005)– Term Extraction

• Shallow Linguistic Parsing & Statistical Analysis– Taxonomy and Relation Extraction

• Shallow Linguistic Parsing & manually defined mapping rules• Named Relations (Subcategorization Frames)



Economic Univ., Prague (Kavalec and Svatek, 2005)– Relation Label Extraction

• Extension of Association Rules Algorithm

Free Univ. Amsterdam (Sabou, 2005)– Term and Taxonomy Extraction (for Web Service Ontologies)

• Shallow Linguistic Analysis & Patterns

Jozef Stefan Inst., Ljubljana -- OntoGen (Fortuna et al., 2005)– Term and Taxonomy Extraction

• Statistical Analysis & Clustering– Relations

• Web Mining, Social Network Analysis

Univ. Paris -- ASIUM (Faure and Nedellec, 1998)– Taxonomy Extraction (& Subcategorization Frames)

• Shallow Linguistic Parsing• Clustering


Univ. Rome – OntoLearn (Navigli and Velardi, 2004; Velardi et al., 2005)– Term Extraction and Interpretation

• Shallow Linguistic Parsing &Term Disambiguation & Compositional Interpretation

– Relations• Classification of the relation between terms in a compound into predefined

set of (thematic) relations– Definitions

• Rules for Gloss Generation

Univ. of Zürich (Rinaldi et al., 2005)– Term and Taxonomy Extraction

• Shallow Linguistic Analysis & Patterns



Terms

Concepts

Taxonomy

Relations

Rules & Axioms

disease, illness, hospital

{disease, illness, Krankheit}

DISEASE:=<Int,Ext,Lex>

is_a(DOCTOR,PERSON)

cure(dom:DOCTOR,range:DISEASE)

(Multilingual) Synonyms

))(),((, xillyxsufferFromyx →∀

Introduced in: Philipp Cimiano, PhD Thesis University of Karlsruhe, forthcoming / also available as Springer book, end of 2006

Ontology Learning Layer Cake


Terms

Concepts

Taxonomy

Relations

Rules & Axioms




is_a(DOCTOR,PERSON)






TermsTerms are at the basis of the ontology learning process

– Terms express more or less complex semantic units– But what is a term?

Huge Selection of Top Brand Computer Terminals Available for Immediate DeliveryBecause Vecmar carries such a large inventory of high-quality computer terminals, including: ADDS terminals, Boundless terminals, DEC terminals,HP terminals, IBM terminals, LINK terminals, NCR terminals and Wyse terminals, your order can often ship same day. Every computer terminal shipped to you is protected with careful packing, including thick boxes. All of our shipping options - including international - are available through major carriers.

– Extracted term candidates (phrases)

- computer- terminal- computer terminal- ? high-quality computer terminal- ? top brand computer terminal- ? HP terminal, DEC terminal, …


Term ExtractionDetermine most relevant phrases as terms

– Linguistic Methods• Rules over linguistically analyzed text

– Linguistic analysis – Part-of-Speech Tagging, Morphological Analysis, …– Extract patterns – Adjective-Noun, Noun-Noun, Adj-Noun-Noun, …– Ignore Names (DEC, HP, …), Certain Adjectives (quality, top, …), etc.

– Statistical Methods• Co-occurrence (collocation) analysis for term extraction within the

corpus• Comparison of frequencies between domain and general corpora

– Computer Terminal will be specific to the Computer domain– Dining Table will be less specific to the Computer domain

– Hybrid Methods• Linguistic rules to extract term candidates• Statistical (pre- or post-) filtering


Terms

Concepts

Taxonomy

Relations

Rules & Axioms




is_a(DOCTOR,PERSON)






Extraction of Synonyms

Term Classification and Clustering

– Classification• Classifying terms to existing class systems, e.g., by

extending WordNet (with SynSets corresponding to classes)

– Clustering• Clusters according to similar distributions, e.g., by

measuring co-occurrence between terms


Terms

Concepts

Taxonomy

Relations

Rules & Axioms




is_a(DOCTOR,PERSON)






The Semiotic TriangleOgden & Richards, 1923

• based on Structural Linguistics studies (de Saussure, 1916)

• adopted in Knowledge Representation (e.g. Sowa, 1984)


Concepts: Intension, Extension, LexiconA term may indicate a concept, if we can define its

– Intension• (in)formal definition of the set of objects that this concept describes

– a disease is an impairment of health or a condition of abnormal functioning

– Extension• a set of objects (instances) that the definition of this concept describes

– influenza, cancer, heart disease, …

Discussion: what is an instance? - ‘heart disease’ or ‘my uncle’s heart disease’

– Lexical Realizations• the term itself and its multilingual synonyms

– disease, illness, Krankheit, maladie, …

Discussion: synonyms vs. instances – ‘disease’, ‘heart disease’, ‘cancer’, …


Concepts: IntensionExtraction of a Definition for a Concept from Text

– Informal Definition• e.g., a gloss for the concept as used in WordNet• OntoLearn (Navigli and Velardi, 2004; Velardi et al., 2005) uses natural

language generation to compositionally build up a WordNet gloss for automatically extracted concepts

– ‘Integration Strategy’ : “strategy for the integration of …”

– Formal Definition• e.g., a logical form that defines all formal constraints on class

membership• Inductive Logic Programming, Formal Concept Analysis, …


Concepts: ExtensionExtraction of Instances for a Concept from Text

– Commonly referred to as Ontology Population– Relates to Knowledge Markup (Semantic Metadata)– Uses Named-Entity Recognition and Information

Extraction

– Instances can be:

• Names for objects, e.g.– Person, Organization, Country, City, …

• Event instances (with participant and property instances), e.g.– Football Match (with Teams, Players, Officials, ...)– Disease (with Patient-Name, Symptoms, Date, …)


Concepts: LexiconExtraction of Synonyms and Translations for a Concept from Text

– (Multilingual) Term Extraction – see previous slides– Representation of Lexical Information in Ontologies

rdfs:subClassOf

rdfs:subClassOfmeta-

classes

classes

instances

rdfs:Class

feat:ClassWithFeats

o:StorageProductfeat:ClassWithFeats

o:Refrigeratorfeat:ClassWithFeats

feat:imgFeatfeat:lingFeat

if:ImgFeat

lf:LingFeat

rdfs:Class

rdfs:Class

lf:lang “de”lf:term “Kühlschrank”lf:morphlf:context

lf:LingFeat

lf:head “Schrank”lf:pos “noun”

lf:Morph

...

if:color “#111111”if:shape “cuboid”lf:texture “&keypatchSet_223”

if:ImgFeat

URIrdf:type

property ...Lege

nd

o:Cupboardfeat:ClassWithFeats

feat:lingFeat

lf:lang “de”lf:term “Schrank”lf:morphlf:context

lf:LingFeat

...

...

...


The Mathematical Definition of an Ontology [Stumme et al.; abbrev. from Cimiano-06]

• Structure:

– C: set of concept identifiers– R: set of relation identifiers– <C partial order on C (concept hierarchy) – <R: partial order on R (relation hierarchy)– Signature:

– Mathematical definition of extension of concepts [c] and relations [r]

– L-Axiom System: Arbitrary Axioms (may include patterns)

+→CR:σ

),,,,(: σRC RCC <<=


LexiconDef: A Lexicon for an ontology is a structure

Lex:={SC,SR,RefC,RefR}

SC,SR are called signs for concepts and relations, respectively.

RefC,RefR, are binary relations denoting lexical referencesfor concepts and relations, respectively.

Example: RefC(„car“)={car-concept1,car-concept2}RefC(„automobile“)={car-concept1}RefC-1(car-concept1)={„car“, „automobile“}


Terms

Concepts

Taxonomy

Relations

Rules & Axioms




is_a(DOCTOR,PERSON)






Distributional Hypothesis & Vector Space Model

• Harris, 1986– „Words are (semantically) similar to the extent to which they share

similar words“• Firth, 1957

– „You shall know a word by the company it keeps“

• Idea: collect context information and represent it as a vector:

• compute similarity among vectors wrt. a measure

XXexcursionXXtrip

XXXXmotor-bikeXXXcar

XXapartment

join_objride_objdrive_objrent_objbook_obj


Context Features• Four-grams [Schuetze 93]

• Word-windows [Grefenstette 92]

• Predicate-Argument relations (every man loves a woman)Modifier Relations (fast car, the hood of the car)– [Grefenstette 92, Cimiano 04b, Gasperin et al. 03]

• Appositions (Ferrari, the fastest car in the world)– [Hahn & Schnattinger 98, Caraballo 99]

• Coordination (ladies and gentlemen)– [Caraballo 99, Dorow and Widdows 03]


Overall Process

Or other clustering mechanism


Using Syntactic Surface Dependencies

Mopti is the biggest city along the Niger with one of the most vibrant ports and a large bustling market. Mopti has a traditional ambience that other towns seem to have lost. It is also the center ofthe local tourist industry and suffers from hard-sell overload. The nearby junction towns of Gao and San offer nice views over the Niger’s delta.

city: biggest(1)ambience: traditional(1)center: of_tourist_industry(1)junction town: nearby(1)market: bustling(1)port: vibrant(1)overload:suffer_from(1)tourist industry: center_of(1), local(1)town: seem_subj(1)view: nice(1), offer_obj(1)


Context Extraction Process• extract syntactic dependencies from text

⇒ verb/object, verb/subject, verb/PP relations⇒ car: drive_obj, crash_subj, sit_in, …

s

LoPar

vpdp

v dp

tgrep

crashed_subj(cars)sat_in(car)drove_obj(car)

sit_in(car)crash_subj(car)drive_obj(car)

lemmatization


Weighting

• Observation:– output of the parser can be erroneous– not all attribute/object pairs are significant

• Conditional Probability:

• Consider attribute/object pairs with weightover threshold t

)|( argvnP


Set Theoretical & Probabilistic Clustering

XXexcursion

XXtrip

XXXXmotor-bike

XXXcar

XXapartment

joinableridabledrivablerentablebookable

• Set theoretical– Formal Concept Analysis

[Ganter and Wille 1999]


Tourism Formal Context

XXexcursion

XXtrip

XXXXmotor-bike

XXXcar

XXappartment

joinablerideabledriveablerentablebookable


Tourism Lattice


Concept Hierarchybookable

rentable joinable

driveable appartment

car

motor-bike

tripexcursion

rideable


Compacting the hierarchybookable

rentable joinable

driveable appartment

carmotor-bike

tripexcursion


Evaluation - Data Sets

• Tourism (118 Mio. tokens):– http://www.all-in-all.de/english– http://www.lonelyplanet.com– British National Corpus (BNC)– handcrafted tourism ontology (289 concepts)

• Finance (185 Mio. tokens):– Reuters news from 1987– GETESS finance ontology (1178 concepts)


Precision/Recall/F-MeasureFCA (Tourism)

0

0,2

0,4

0,6

0,8

1

1,2

0 0,2 0,4 0,6 0,8 1

threshold t

PrecRecallF


Lexical Recall, F‘

FCA (Tourism)

00,050,1

0,150,2

0,250,3

0,350,4

0,450,5

0 0,2 0,4 0,6 0,8 1

threshold t

FLRF'


Comparison (Tourism, F‘)

Comparison (Tourism)

00,05

0,10,15

0,20,25

0,30,35

0,40,45

0,5

0 0,2 0,4 0,6 0,8 1

threshold t

FCAComplete LinkageAverage LinkageSingle LinkageBi-Section-Kmeans


Comparison (Finance, F‘)

Comparison (Finance)

00,05

0,10,15

0,20,25

0,30,35

0,40,45

0 0,2 0,4 0,6 0,8 1

threshold t

FCAComplete-LinkageAverage LinkageSingle LinkageBi-Section-Kmeans


Clustering – Comparison

Weak-FairO(n2)36.42/32.77%DivisiveClustering

FairO(n2 log(n))O(n2)O(n2)

36.78/33.35%36.55/32.92%38.57/32.15%

AgglomerativeClustering

GoodO(2n)(pract. better!)

43.81/41.02%FCA

UnderstandabilityWorst Case Time Complexity

F-Measure


Problem 1: Labeling of Clusters• Caraballo’s Method [1999]:

– Agglomerative Clustering– Labeling Clusters with hypernyms derived from

Hearst patterns– Removing unlabeled concepts thus compacting the

hierarchy

• Evaluation: select 20 nouns with at least 20 hypernyms and present them to human judges with the 3 best hypernyms for each

• Results: – Best Hypernym (33% (Majority) / 39% (Any)– Any Hypernym (47.5% (Majority) / 60.5% (Any))


Problem 2: Spurious Similarities• Guided Clustering [Cimiano 2005c]:

– Integrate a externally derived hypernym oracle into the agglomerative clustering algorithm

– Two terms are only clustered if they have a common hypernym according to the oracle

– Label the cluster with the common hypernym⇒Demonstrably better hierarchies⇒Labels for the cluster

⇒Reuse techniques from Clustering with constraints!


Conclusion about Comparison

• FCA is an interesting alternative to similarity-based clustering approaches– high traceability due to intensional description

of clusters– Problem: worst case exponential in the size of

the formal context– But: Zipfian distribution of attributes


Using Ontologies withText Retrieval


Using Ontologies

Ontologies as:

• background knowledge for text clustering and classification

• basis for recommender systems• background knowledge in ILP• knowledge for models in Statistical

Relational Learning


Text Clustering & Classification Approaches

clustering/classification

algorithm

DocumentsBag of Words

backgroundknowledge

oman has granded …Obj1 2 2Obj2 1 1Obj3 2 …Obj4 2 …

1 …0 …

0 00 0


Omanhas grantedtermcrudeoilcustomersretroactivediscounts...

211112111...

Bag of WordsDok 17892 crude ============= Oman has granted term crude oil customers retroactive discounts from official prices of 30 to 38 cents per barrel on liftings made during February, March and April, the weekly newsletter Middle East Economic Survey (MEES) said. MEES said the price adjustments, arrived at through negotiations between the Omani oil ministry and companies concerned, are designed to compensate for the difference between market-related prices and the official price of 17.63 dlrs per barrel adopted by non-OPEC Oman since February. REUTER

Documents

Further preprocessing steps-Stopwords-Stemming

Text Clustering & Classification Approaches


109377 Concepts(synsets)

WordNet as an example and ontology

144684 lexicalentries

Rootentity

something

physical object

artifact

substance

chemicalcompound

organiccompound

lipid

oil

EN:oil

covering

coating

paint

oil paint

cover

cover with oil

bless

oil, anoint

EN:anoint EN:inunct

oil colorcrude oil

144684 lexicalentries

Use of superconcepts(Hypernyms in Wordnet)

• Exploit more generalized concepts• e.g.: chemical compound is the 3rd superconcept of oil

Strategies:all, first, context


Ontology-based representation

strategy: add

Omanhas grantedtermcrudeoilcustomersretroactivediscounts...

111111111...

1

1111111111...

Omangrantedterm(C) termcrude(C) crudeoil(C) oilcustomer(C) customer...

2

1111111111...

3

Omangrantedterm(C) termcrude(C) crudeoil(C) oil(C) lipid(C) compound...


Evaluation of Text Clustering0,6160,618

0,570

0,300

0,350

0,400

0,450

0,500

0,550

0,600

0,650

add repl add only repl add only repl add only repl add only repl add only repl add only

context context first all context first all

0 0 5

false true

tfidf - 30without - 30

CLUSTERCOUNT 60 EXAMPLE 100 MINCOUNT 15

Mittelwert - PURITY

ONTO HYPDEPTH HYPDIS HYPINT

WEIGHT

PRUNE

backgro..depthdisambig.integrat.

Evaluation parameter• min 15, max 100, 2619 documents

of the reuters corpus• cluster k = 60, with BiSec-KMeans

avg - purity


Evaluation: OHSUMED Classification Results

Top 50 classes with WordNet and AdaBoost


Combine FCA & Text-clustering

1. preprocess Reuters documents and enrich them with background knowledge (Wordnet)

2. calculate a reasonable number k (100) of clusterswith BiSec-k-Means using cosine similarity

3. extract a description for all clusters4. relate clusters (objects) with FCA5. use the visualization of the concept lattice for

better understanding


Explaining Clustering Results with FCA

compound, chemical compound

oil

refiner

chain of concepts with increasing specificity



Crude oilbarrel



resin palm

• Resulting concept latticecan also be interpretedas a concept hierarchydirectly on thedocuments

• all documents in onecluster obtain exactlythe same description


Conclusion: Ontologies + Text

• Ontologies may be discovered as regularities unterlying some text

• Ontologies improve access to text– By annotation (cf part 2)– By retrieval (this part)