Wissenstechnologie Iii 08 09

Wissenstechnologie WS 08/09

Michael Granitzer

IWM TU Graz & Know-CenterIWM TU Graz & Know Center

http://kmi tugraz at http://www know center athttp://kmi.tugraz.at http://www.know-center.atThis work is licensed under the Creative Commons Attribution 2.0 Austria License. To view a copy of this license, visit http://creativecommons.org/licenses/by/2.0/at/.

TodayToday

h bThe Semantic Web Stack (rep )Stack (rep.)

Semantics & Semantics & Ontologies

RDF S h (RDFS)2

RDF Schema (RDFS)

WS 08/09

http://kmi.tugraz.at

Wissenstechnologie @ kmi.tugraz.at

Definition „Semantic Web“The Semantic Web Stack (rep.)

The Semantic Web is an extension of the current Web in The Semantic Web is an extension of the current Web in which information is given well-defined meaning, betterenbaling computers and people to work in cooperations.

[Berners-Lee et al. 2001]

http://www.sciam.com/print_version.cfm?articleID=00048144-10D2-1C70-84A9809EC588EF2110D2 1C70 84A9809EC588EF21

3

WS 08/09



The Vision as Application ScenarioThe Semantic Web Stack (rep.)

Plan a trip via the internet using your personal agentPlan a trip via the internet using your personal agent

Agent searches automatically for

Suitable flightSuitable flight

Suitable hotels

Alternative routesAlternative routes

Also, the software agent tells you why it made this decision!

4

WS 08/09



How to Express Semantics The Semantic Web Stack (rep.)

A small example

John Lennon

Th B tlBandIs Member Is A

The Beatles

Is MemberPaul McCartney

Query: all bands from EnglandIs born in

Founded in

Liverpool England

Query: all bands from England

?All bands with English artists?

Is born in

Ist in

5

Liverpool England ?All bands with English artists?

Inferenz & Reasoning: E li h i P h i i d b i E l d

WS 08/09



English artists := Person who is an artist and born in England

Semantic Web StackThe Semantic Web Stack (rep.)

a.k.a. SW Layer Cakey

a.k.a. SW Tower

6

WS 08/09




Unicode

URI

7

WS 08/09




XML

XML Schema

Namespaces

8

WS 08/09



Drawbacks of XMLThe Semantic Web Stack (rep.)

9

WS 08/09



Drawbacks of XMLThe Semantic Web Stack (rep.)

No semantic/meaning of tags No semantic/meaning of tags

Tree-like structure makes it hard to combine decentralstored information

<Person>

<name> x</name>

<lecture>

/<name> x</name>

<lecture>

…</lecture>

<name> x</name>

<Person>

…/

10

</lecture>

</Person>

</Person>

</lecture>

WS 08/09




RDF

11

WS 08/09



Goal of RDFThe Semantic Web Stack (rep.)

Description of (Web) resource via metadataDescription of (Web) resource via metadata

Historically focused on web sites

E t d d t l“ Extended to „general“ resources

For

Classification of resources

Classification of relationships between resources

Unambigious description

12

Unambigious description

WS 08/09



RDF Statements (Triples)The Semantic Web Stack (rep.)

A small example

htt // iki di / iki/J h L http://dbpedia org/property/associatedActshttp://en.wikipedia.org/wiki/John_Lennon

http://en.wikipedia.org/wiki/The_Beatles

http://dbpedia.org/property/associatedActs

http://en.wikipedia.org/wiki/Paul_McCartney

rdfs:label


Subject Predicate Object

„Paul McCartney“

j j

http://en.wikipedia.org/wiki/John_Lennon



13http://en.wikipedia.org/wiki/Paul_McCartney



http://en.wikipedia.org/wiki/P Rdfs:label “Paul McCartney”

WS 08/09



aul_McCartney

RDF – SerialisationTurtle Example - Extended

The Semantic Web Stack (rep.)

Turtle Example - Extended

# Define some namespaces

@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .

@prefix dc: <http://purl.org/dc/elements/1.1/> .

@prefix ex: <http://example org/terms/>@prefix ex: <http://example.org/terms/> .

<http://www.example.org/index.html>

dc:creator <http://www.example.org/staffid/85740> .

# write all statements in short form

<http://www.example.org/staffid/85740>

ex:name "John Smith";

ex:age "27" .

14

WS 08/09



RDF Extended ConceptsThe Semantic Web Stack (rep.)

Blank NodesBlank Nodes

Container & Collections

ReificationReification

Syntactical abbreviations, no extension of expressiveness

But how to define meaning?

15

WS 08/09



Ontologies & SemanticsWhat is an Ontology?

Semantics & Ontologies

What is an Ontology?

Greek: The study of being“Greek: „The study of being

Branch of Philosophy

W it d t th d fi iti f t i We can narrow it down to the definition of concepts in the world and their relationship

16

WS 08/09



OntologiesWhat are Concepts in our purpose?


What are Concepts in our purpose?

Semiotic Triangle [Ogden & Richards 1923]Semiotic Triangle [Ogden & Richards 1923]

Concept

Refers toSymbolizes

Term / Word/URI

St d fThing

17‚Apache‘

Stands for

WS 08/09



Ontologies & SemanticsHow to describe concepts?


How to describe concepts?

Intensional Description: Conditions and properties of a Intensional Description: Conditions and properties of a concept

Natural World: textual summaryy

Logics:

N d ffi i di i– Necessary and sufficient conditions– constraints on things

Extensional Description: List of all objects belonging to a p j g gconcept

18

WS 08/09



Ontologies & SemanticsExample: Mammal


Example: Mammal

IntensionIntension

•isA(Vertebrate Animal)•has(Sweat glands)

•withFunction(Milk)•withFunction(hair)

•....

Extension

•Elephant•Lion•Monkey

19

Monkey•....

WS 08/09



Ontologie (Gruber)Semantics & Ontologies

Definition in Computer Science

explicit specification of a conceptualization

conceptualization is an abstract, simplified view ofp , pthe world that we wish to represent for some purpose

Definitions associate the names of entities in theuniverse of discourse with human readable textuniverse of discourse with human-readable textdescribing what the names mean, and formal axiomsthat constrain the interpretation and well-formed useof these terms of these terms. Formally, an ontology is the statement of a logicaltheory

20

WS 08/09



Ontologie (Gruber)


Ontologie (Gruber)

Ontologies are often equated with taxonomic Ontologies are often equated with taxonomic hierarchies of classes, but class definitions, and the subsumption relation, but ontologies need not be limited to these forms To specify a limited to these forms. … To specify a conceptualization one needs to state axioms that do constrain the possible interpretations for the d fi d tdefined terms.

21

WS 08/09



Ontologie (Guarino)Semantics & Ontologies

Language vs. Conceptualization

An ontology is a logical theory accounting for the gy g y gintended meaning of a formal vocabulary, i.e. its ontological commitment to a particular conceptualization of the world. The intended models of a logical language using such a vocabulary are constrained by its ontological commitment. An ontology indirectly reflects this commitment (and the underlying conceptualization) by

h d d d lapproximating these intended models.

an ontology is language-dependent

22a conceptualization is language-independent

WS 08/09



Ontologie (Sowa)Semantics & Ontologies

Formalization level of Ontologies

An informal ontology may be specified by a catalog of types that are either undefined or d fi d l b t t t i t l l defined only by statements in a natural language.

A formal ontology is specified by a collection of names for concept and relation types organized names for concept and relation types organized in a partial ordering by the type-subtype relation.

23

WS 08/09



Ontologie (Obrst)


Ontologie (Obrst)

With respect to definitions of ontologies, I hope to send a portion of a briefing I made at the Army Knowledge Management Conference in Ft. Lauderdale late Aug/early Sept of 2004, that takes you through the ontology spectrum from taxonomy (weak takes you through the ontology spectrum, from taxonomy (weak and strong) to thesaurus (a strong term taxonomy) to conceptual model (weak ontology) to logical theory (strong ontology).

The first is unstandardized the second and third each has a The first is unstandardized, the second and third each has a set of standards associated with them, the third and fourthhave multiple representation languages supporting them, and the last has some logic behind the representation language, typically ranging from a description logic (OWL) to first-order typically ranging from a description logic (OWL) to first order logic (KIF, Common Logic) to a higher order logic.

A logical theory is a formal ontology. The others range from informal to semi-formal. Other informal ontologies can be natural language sentences in a document The key point

24

natural language sentences in a document. The key point about formal ontologies (logical theories) is that they are machine-interpretable, i.e., semantically interpretable by machine. The others are not, are only interpretable by human beings, though they may be machine-readable and

WS 08/09



human beings, though they may be machine readable andmachine-processable.

Summary of DefinitionsSemantics & Ontologies

A Ontology is a model (of the world)

A t l d ib ti l (k l d ) d iA ontology describes a particular (knowledge) domain

A ontologie defines words/terms/signs for describingConceptsConcepts

A ontologie puts concepts into relation to each other

A ontologie uses axioms to put constraints on particularconcepts

25

WS 08/09



Components of an Ontology


Components of an Ontology

Classes general things of a domainClasses general things of a domain

Instances special things of a domain

R l ti b t thiRelations between things

Properties of things

26

WS 08/09



Semantics & CommunicationWhy do we need Ontologies in the Web?


Java based Intelligent Agent

C# basedIntelligent AgentExchange Semantics

on the basis of an

Q: Is Paul McCartney member of a Rock Band?

agreed Ontology

27

WS 08/09



Semantics & CommunicationSemantics & Ontologies

Language must allow to express the semantics in an Language must allow to express the semantics in an implementation/algorithmic independent way

Usually done via a Vocabulary

Topic oriented vocabulary (e.g. Friend of a friend)

Schema Knowledge/Terminological Knowledgeg g g

– Special vocabulary to make statements over topic orientedvocabulary (i.e. the termonologie used in a domain)

– A general set of rules independent of the domain– Defines the expressiveness of a language

28

WS 08/09



Semantics & Communication Example


Example

Topic Vocabulary: Elephant, Mammal, AnimalTopic Vocabulary: Elephant, Mammal, Animal

Schema: isSubClassOf defines an transitiv IS-A relationshipisSubClassOf defines an transitiv IS A relationship

Define that: isSubClassOf(Elephant, Mammal)==true

Define that: isSubClassOf(Mammal, Animal)==true

isSubClassOf(Elephant,Animal)==true

Independent of implementation and applyable toabritrary vocabularies:abritrary vocabularies:

isSubClassOf(A, B)

isSubClassOf(B, C)

29

isSubClassOf(B, C)

isSubClassOf(A,C)==true

WS 08/09



Semantics & Communication Example


Example

Similar „Rules“ exist in natural languageSimilar „Rules exist in natural language

Fact 1: „An elephant is a mammal“„Mammals like for example elephants“

Fact 2: „A mammal is an animal“

Based on our formal knowledge we conclude that an g„elephant is an animal“.

Note: Exploitable in Ontology Learning from Text

30

WS 08/09



Ontology Spectrum (McGuinness)Semantics & Ontologies

..or how much semantic expresses

SelectedLogical

Thesauri“narrower

t ”Formal

iFrames

(properties)Catalog/ID

LogicalConstraints

(disjointness, inverse, …)

term”relation

is-a (properties)

inverse, …)

Terms/Informal

is-aFormal

instance Value GeneralLogical

glossary Restrs. constraints

31Originally from AAAI 1999- Ontologies Panel by Gruninger, Lehmann, McGuinness, Uschold, Welty;– updated by McGuinness.

http://ontolog.cim3.net/file/work/OntologySummit2007/workshop/McGuinness_NIST-interop-ontology-summit_20070423.ppt

WS 08/09



Description in: www.ksl.stanford.edu/people/dlm/papers/ontologies-come-of-age-abstract.html

Semantic Web StackRDF Schema (RDFS)

RDF Schema

32

WS 08/09



RDF Schema (RDFS)http://www w3 org/2000/01/rdf-schema#

RDF Schema (RDFS)

http://www.w3.org/2000/01/rdf-schema#

Allows to express terminological knowledge over RDFAllows to express terminological knowledge over RDF

Application of RDFS

Defines a new vocabulary for giving meaningDefines a new vocabulary for giving meaningindependent of program logic

Allows to define „lightweight“ Ontologies and basicg g gReasoning capabilities

http://www.w3.org/TR/rdf-schema/

33

WS 08/09



RDF Schema & Object-Orientierted

RDF Schema (RDFS)

Languages

RDFS uses object-oriented Concepts:j p

Classes

Properties of the classes

But not classes have properties (e.g. Java)

Properties are assigned to classes:

Easier to extend vocabularyEasier to extend vocabulary

Easier to assign properties to classes

Take care on uniqueness of Propertiesq p

34

WS 08/09



RDF SchemaNotation

RDF Schema (RDFS)

Notation

@prefix rdfs <http://www w3 org/2000/01/rdf-schema#>@prefix rdfs <http://www.w3.org/2000/01/rdf schema#>.

@prefix rdf <http://www.w3.org/1999/02/22-rdf-syntax-ns#>.

For the following slides we define this namespaceFor the following slides we define this namespace

35

WS 08/09



RDF SchemaClasses

RDF Schema (RDFS)

Classes

rdfs:Resource Class of all resources

rdfs:Literal Class of literals (Strings)rdfs:Literal Class of literals (Strings)

rdf:XMLLiteral Class of XML Literals

rdfs:Class Class of classesrdfs:Class Class of classes

rdf:Property Class of properties

rdfs:Datatype Class of datatypes (e g integer etc )rdfs:Datatype Class of datatypes (e.g. integer etc.)

rdf:Statement Class of RDF Statements

rdfs:Container Class of containers

36

rdfs:Container Class of containers

WS 08/09



RDF SchemaProperties

RDF Schema (RDFS)

Properties

rdf:type Subject is an instance of a class

rdfs:subClassOf Subject is a subclass of a class

rdfs:subPropertyOf Subject is a sub property of a property

rdfs:domain A possible class for a subject of a property

rdfs:range A possible class for an object of a property

rdfs:label human readable label of an resource

rdfs:comment human readable comment of an resourcerdfs:comment human readable comment of an resource

…

37

WS 08/09



RDF SchemaInstances Classes

RDF Schema (RDFS)

Instances, Classes

Typing: Individuals are assigned to classes (multiple Typing: Individuals are assigned to classes (multiple assignments possible)

rdfs:Class

rdf:type

#Car

#MyBMW

rdfs:subClassOfrdf:type

rdfs:Resource

38Note: Sometimes it is domain dependent what an instance isand what not (modelling aspect)

WS 08/09



RDF SchemaHierarchies

RDF Schema (RDFS)

Hierarchies

rdf:subClassOf allows to define hierarchies amongclasses

#Means of Transportation

#Electric vehicle

#MyBMW

rdfs:subClassOf

39

#Car

rdfs:subClassOf#BMW

rdf:type#Train

WS 08/09



RDF SchemaHierarchies

RDF Schema (RDFS)

Hierarchies

Rdf:subPropertyOf allows to define hierarchies amongproperties

ex:has

– ex: hasFourex: hasFour– ex:hasTwo

<#BMW>ex:hasFour <#Tires> .

<#BMW>ex:has <#Tires> .

40

ex:has <#Tires> .

WS 08/09



RDF SchemaDomain & Range

RDF Schema (RDFS)

Domain & Range

rdf:Domain and rdf:Range allow to specify whichclasses of subjects (==domain) and which classes ofbj t ( ) t tobject (==range) a property can connect

<ex:has> rdf:domain <#Car>

<ex:has>rdf:Range <rdf:Resource>

41

WS 08/09



RDF SchemaExampleExample

42

WS 08/09



RDFS SemanticsRDFS Semantics

Model-theoretic semantics (subfield of formal semantics)Model theoretic semantics (subfield of formal semantics)

Entailment: Given a graph the graph is transformed accordingto the rules of RDFS

Implicit knowledge (i.e. not explicitly modelled)


rdfs:subClassOf


rdfs:subClassOfrdf:type

#Car

rdfs:subClassOf

#MyBMW

rdf:type#Car

df bCl Of

#MyBMW

df

rdfs:subClassOfyp

43

rdfs:subClassOf

#BMW

rdfs:subClassOf

#BMW

rdf:type

WS 08/09



RDFS SemanticsDeductive Rules/Entailment

RDFS Semantics

Deductive Rules/Entailment

The RDF Semantics Document defines a list of 44 Entailment The RDF Semantics Document defines a list of 44 Entailment Rules:

ddlidif1 nss K

“do that recursively until the graph does not change”

sstatement add,statementsvalidare if 11

nn ss

sK

do that recursively until the graph does not change

“this can be done in polynomial time for a specific graph”

We have means for how statements should be interpreted

W “ i ” f URI’ i RDFS

44

We can express “meaning” of URI’s using RDFS

http://www.w3.org/TR/rdf-mt/

WS 08/09



RDFS SemanticsEntailment Example

RDFS Semantics

Entailment Example

u x v URI‘s or Blank Nodesu, x, v …. URI s or Blank Nodes

.:: Classrdfssubtyperdfsu.Re:: sourcerdfssubClassOfrdfsu

rdfs:subtyperdfs:Class

rdfs:Resource#Car rdfs:subClassOf

.:

.:.:xsubClassOfrdfsu

xsubClassOfrdfsvvsubClassOfrdfsu #Means of Transportation

rdfs:subClassOf

45#Car

rdfs:subClassOf

WS 08/09



#BMW

RDFS SemanticsDrawback/Restriction of RDF

RDFS Semantics

Drawback/Restriction of RDF

Open world assumption: false statements must be Open world assumption: false statements must be specified

Closed world assumption: if a statement is missing, it is p g,assumed to be false

No negation in RDFS possible

• ex:michael rdf:type ex:nonsmoker

• ex:michael rdf:type ex:smoker• ex:michael rdf:type ex:smoker

Does not lead to a contradiction!

N l i di id l H All

46

No rules over individuals e.g. ex:Humans = All ex:Women and All ex:Men

No Counting: “An Elephant has 4 legs”

WS 08/09



No Counting: An Elephant has 4 legs

Summary

Ontology = Classes Instances Properties andOntology = Classes, Instances, Properties andRelationships

RDFS as terminological vocabulary over RDFg y

RDF Schema (RDFS):

First step in increasing semanticsFirst step in increasing semantics

No negation and restricted logic capabilities

47

WS 08/09



Points you should take away from thislecturelecture

• What are Ontologies in Computer Science?• What are Ontologies in Computer Science?

• What adds RDFS to the semantic expressiveness of RDF

Wh i RDFS t h?• Why is RDFS not enough?

48

WS 08/09



That‘s it for today…

Thanks for your attention

Questions/comments?

i @[email protected]

49

WS 08/09



License

This work is licensed under the Creative Commons This work is licensed under the Creative Commons Attribution 2.0 Austria License. To view a copy of this license, visit http://creativecommons org/licenses/by/2 0/at/http://creativecommons.org/licenses/by/2.0/at/.

Contributors:

Mathias Lux

Peter Scheir

Klaus Tochtermann

50Michael Granitzer

WS 08/09



Wissenstechnologie Iii 08 09

Education

d http

semantic web stackrdf11

mccartney http

semantic web stack rep

beatles http

orgwikip http

orgwikij http

definition semantic