Knowledge Representation - h da Hochschule Darmstadt

54
Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019 1 Knowledge Representation Applied Artificial Intelligence Prof. Dr. Bernhard Humm Faculty of Computer Science Hochschule Darmstadt – University of Applied Sciences

Transcript of Knowledge Representation - h da Hochschule Darmstadt

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.20191

Knowledge RepresentationApplied Artificial Intelligence

Prof. Dr. Bernhard HummFaculty of Computer ScienceHochschule Darmstadt – University of Applied Sciences

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.20192

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

AI landscape

3

Non-symbolicAI /

Machinelearning

Knowing

Knowledge representation

Learning

Machine learning, Information retrieval,

Data mining

Symbolic AI /Knowledge-

based AI

Reasoning

Logic programming,Probabilistic reasoning

Complex event processing

Communicating

Natural language processing

Perceiving

Computer vision,Sensor technology

Acting

Planning, Agent technology,

Robotics

Applyingintelligence

Acquiringintelligence

Inspired by https://www.sigs-datacom.de/order/poster/Landkarte-KI-Poster-AI-2018.php

• Semantic networks• Ontologies• Logic programming

• Artificial neural networks• Support vector machines• Linear / logistic regression• Random forest• Gradient boosting tree• K-Nearest neighors• K-means• Naive Bayes

• Bayes networks• Hidden Markov models• Decision tree learning• Inductive logic programming

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Knowldege to be representedExample: Arts

• Michelangelo was an Italian artist

• He created many paintings, e.g., the

„Creation of Adam“ in the Sistine Chapel, Rome

• He also created sculptures like David

• He belonged to the artistic movement of

High Renaissance

• People who create paintings are painters

• Painters are artists

4 Image source: wikimedia

Individuals Classes

Relationships Rules

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Ontology

• Knowldege represented for a particular application domain,

e.g., arts

• Contains:

- Schema: class types (e.g., “person“), relationship types (e.g., “has

painted“, and rules

- Facts: individuals (class instances, e.g., “Michelangelo“) and their

relationships (e.g, “Michelangelo painted ‘Creation of Adam‘“)

• Is formalized with a particular knowledge reprentation approach / language

5

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Knowledge Representation approaches

•painted (Michelangelo, CreationOfAdam)

•p: ( x: painted (p, x)) painter (p)

PredicateLogic

Frames

Semantic Nets

Rules

6

Michelangelo

Type: Person

born: 1475

in: Italy

CreationOfAdam

type: Painting

artist:Michelangelo

(?p instance-of person)

(?p painted ?x)

-->

(?p instance-of painter)

Michelangelo

CreationOfAdam

Italy

created

born-in

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Knowledge representationand reasoning

7

Known (assertedfacts and rules)

• Michelangelo is a person

• Michelangelo has created the painting „Creation of Adam“

• Every person that has created a painting is a painter

Derived(via reasoning)

• Michelangelo is a painter

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Knowledge Representation and Querying

Who has painted„Creation of

Adam“?Michelangelo

Has Michelangelo painted „Creation

of Adam“?Yes

Find all paintersMichelangelo,

9

Requiresreasoning!

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201910

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Resource Description Framework (RDF)

• Knowledge representation language

• Partially based on predicate logic and semantic nets

• Specified by World Wide Web Consortium (W3C) as part of the

Semantic Web initiative

http://www.w3.org/standards/techs/rdf#w3c_all

• Further W3C specifications:

- RDF Schema (RDFS)

- Web Ontology Language (OWL)

11

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Resources and URIs

• Resource: anything that someone might want to talk about, e.g.

- Michelangelo

- All painters

- Property “created”

• Every resource has a Uniform Resource Identifier (URI), e.g.

12

<https://www.wikidata.org/entity/Q5592>

Michelangelo

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Namespaces

• Abbreviation scheme:

qnames with namespaces

13

@prefix wd: <http://www.wikidata.org/entity/> .

wd:Q5592

Michelangelo.Abbreviation for <https://www.wikidata.org/entity/Q5592>

Namespace prefix definition

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Base namespace

• One namespace can be defined as @base

• This namespace can be used without prefix

14

@base <http://h-da.de/fbi/artontology/> .

:artwork

Abbreviation for <http://h-da.de/fbi/artontology/artwork>

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Some common namespacesName-space Description URI

rdf: ResourceDescription Framework

http://www.w3.org/1999/02/22-rdf-syntax-ns#

rdfs: RDF Schema

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

owl: Web OntologyLanguage

http://www.w3.org/2002/07/owl#

xsd: XML Schema

http://www.w3.org/2001/XMLSchema#

dbpedia: DbpediaResource

<http://dbpedia.org/resource/>

wd Wikidata <http://www.wikidata.org/entity/>

yago YAGOOntology

<http://dbpedia.org/class/yago/>

15

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Triples:The basic statement notation of RDF

• Statement including subject, predicate, and object

• Subject: resource

Predicate: resource

Object: resource or literal value (XML data type)

• Examples:

16

wd:Q5592 rdfs:label "Michelangelo" .

wd:Q5592 rdf:type :person .

wd:Q5592 :date_of_birth 1475 .

wd:Q5592 :date_of_death 1564 .

wd:Q5592 :movement wd:Q1474884.

Subject: Resource Predicate: Resource

Object: String Literal (English)

Object: Literal or resource

wd:Q5592

"Michelangelo"

1475

wd:Q1474884

rdfs:label

:date_of_birth

:movement

Triple terminates with full stop

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Abbreviation for mulitple triples withthe same subject: ;

17

wd:Q5592 rdfs:label "Michelangelo" ;

rdf:type :person ;

:date_of_birth 1475 ;

:date_of_death 1564 ;

:movement wd:Q1474884.

Semicolon chains multiple triples with identical subject

wd:Q5592

"Michelangelo"

1475

wd:Q1474884

rdfs:label

:date_of_birth

:movement

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Classes and Instances (Individuals):rdf:type and rdfs:Class

• rdf:type expresses instance relationship

• rdfs:Class represents a meta-class (case-sensitive, not rdfs:class!)

• Examples:

• Abbreviation for rdf:type : a

18

:person rdf:type rdfs:Class .

wd:Q5592 rdf:type :person .

:person a rdfs:Class .

wd:Q5592 a :person .

„Michelangelo is a person“Is-a relationship between individual and class

„Person is a class“ Schema information (class definitions) with thesame language constructs as facts.

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Properties

• rdf:Property

- Similar to rdfs:Class

- Declares a Property that can be used as a predicate

- Example:

19

:movement rdf:type rdf:Property .

wd:Q5592 :movement wd:Q1474884 .

Movement is a property

Michelangelo High Renaissance

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201920

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Linked open data

• Large set of ontologies

• Based on RDF

• Publicly available

• Numerous domains, e.g.

life sciences, geography,

media, social networking

• Examples: dbpedia,

YAGO, MeSH, Wikidata, …

21

https://lod-cloud.net

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Wikidatahttps://www.wikidata.org

• Free knowledge base as the basis for Wikipedia info boxes

• Accessible via SPARQL endpoint https://query.wikidata.org/

22

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Example: Mona Lisahttps://www.wikidata.org/wiki/Q12418

23

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

ArtOntology: Extract of Wikidata

24

90,000 artworks

26,000 artists, e.g. Leonardo da Vinci

210 movements

e.g. Renaisance

230 genrese.g.

portrait

4.300 locations

e.g. Louvre

18,000 motifs

e.g. mountain

460 materials

e.g. oil paint

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

ArtOntology Schema

25

:artwork

+id : URI+rdfs:label : string+:description : string+:image : URL+:inception : int+:country : string+:height : number+:width : number

:person

+id : URI+rdfs:label : string+:description : string+:image : URL+:gender : string+:date_of_birth : int+:date_of_death : int+:place_of_birth : string+:place_of_death : string+:citizenship : string

:movement

+id: URI+rdfs:label : string+:description : string+:image : URL:genre

+id : URI+rdfs:label : string+:description : string+:image : URL

:material

+id : URI+rdfs:label : string+:description : string+:image : URL

:object

+id : URI+rdfs:label : string+:description : string+:image : URL

:location

+id : URI+rdfs:label : string+:description : string+:image : URL+:country : string+:website : URL+:lat : number+:lon : number

+:material

+:genre

+:movement +:creator

+:location

+:depicts

+:movement

+:influenced_by

+:part_of

+:influenced_by

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

File ArtOntology.ttl

26

~50 MBDownload from my home page

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201927

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

SPARQL Query Language

• Query Language for RDF

• Standardized by W3C

• Current Version: 1.1

• Specification: http://www.w3.org/TR/sparql11-query/

28

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Namespaces

• Namespace prefix definition (slightly different from RDF)

• Example:

29

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

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

PREFIX wd: <http://www.wikidata.org/entity/>

PREFIX : <http://h-da.de/fbi/artontology/>

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Simple SELECT query: „What is theplace of birth of Paul Gauguin?“

30

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

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

PREFIX wd: <http://www.wikidata.org/entity/>

PREFIX : <http://h-da.de/fbi/artontology/>

SELECT ?c

WHERE {

wd:Q37693 :place_of_birth ?c .

}

Query variable prefixed by ‘?‘

Keywords SELECT, WHERE simliar to SQL

Result when executed in Apache Fuseki:“Paris“

Query constraints as RDF triples withquery variables

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

And if we don‘t know the URI of Paul Gauguin?

31

SELECT ?c

WHERE {

?p rdf:type :person ;

rdfs:label "Paul Gauguin" ;

:place_of_birth ?c .

}

AbbreviatedNotation as in RDF

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Which other persons were born in Paris?

• No

32

SELECT ?l

WHERE {

?p rdf:type :person ;

rdfs:label ?l ;

:place_of_birth "Paris" .

}

Oh la la,600 artists from

Paris in theontology

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Multiple result variables:„Artists with their place of birth“

• No

33

SELECT ?l ?pob

WHERE {

?p rdf:type :person ;

rdfs:label ?l ;

:place_of_birth ?pob .

}

LIMIT 1000

Multiple queryvariables

One column for eachquery variable in the

result set

We better limit theresult size

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

SELECT *All query variables are returned

34

SELECT *

WHERE {

?p rdf:type :person ;

rdfs:label ?l ;

:place_of_birth ?pob .

}

LIMIT 1000

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

SELECT DISTINCT:“Which are the countries of artworks?“

• No

35

SELECT DISTINCT ?c

WHERE {

?p rdf:type :artwork ;

:country ?c .

}

Duplicates are removed fromthe result set

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Path expressions:“Artworks of Paris-born artists“

• No

36

SELECT ?n ?l

WHERE {

?a rdf:type :artwork ;

rdfs:label ?l ;

:creator/:place_of_birth "Paris" ;

:creator/rdfs:label ?n .

}Path expression using ‘/‘.

Abbreviation for:?a :creator ?c .

?c :place_of_birth "Paris" .

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Transitive closures: “Persons who wereinfluenced by Pieter Brueghel the Elder“

• No

37

SELECT *

WHERE {

?p rdf:type :person ;

:influenced_by* wd:Q43270 ;

rdfs:label ?l .

}‘*‘: Zero to n predicates :influenced_by in sequence, e.g.

Pieter Brueghel the Elder -> Peter Paul Rubens -> Vincent van Gogh -> Wassily Kandinski.

‘+‘: 1 to n predicates in sequence

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

More SPARQL concepts

• ASK queries: checks for a condition and returns a boolean result

• CONSTRUCT queries: generate new RDF statements

like production rules

• FILTER: allows to formulate additional conditions, e.g., on datatypes

• OPTIONAL: specifies optional values

• EXISTS / NOT EXISTS: negation

• GROUP BY, HAVING: aggregation

• ORDER BY: sorting

• Subqueries

• …

• For details see http://www.w3.org/TR/sparql11-query

38

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201939

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Open vs. Closed World Assumption

• Are these all? Are you sure?

• Closed Word Assumption (CWA)

- The database (asserted and inferred facts) is complete

- Everything that cannot be found or inferred is not existing (wrong)

- Negation as failure, e.g. „Is Rembrandt a sculptor?“ „No, since

we don‘t know a sculpture of his“

• Open World Assumption (OWA)

- Facts can be added to the database at all times

- „Never say never“

40

Find all painters

Michelangelo, Raffael, Leonardo da Vinci, Dürer, Rembrandt, Picasso, Magritte.

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Discussion CWA versus OWA

• Assume an arts ontology with the facts and rules as before.

Answer the following under the CWA and the OWA:

1. Who painted „Creation of Adam“?

2. Is Michelangelo an Artist?

3. Which paintings did Michelangelo create?

4. How many paintings did Michelangelo create?

• Discussion

- Who decides between CWA and OWA?

- What ist the technical implication of the decision between CWA

and OWA?

- What is the business implication? How to interpret query results

under CWA and OWA?

41

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Classes / Instances in RDF(S)vs. Object-Orientation

• Concept of class (group) and instances (individuals) in RDF(S)

comparable with Object-Orientation

• But there are major differences:

42

Object-Orientation Semantic Web

No instance without class Any resource can be used – no needfor specifying a class and an rdf:type relationship

Instance belongs to exactly one class Zero, one, or many rdf:typerelationships may be specified

Class specifies attributes and methods of all instances

No restrictions on the relationshipsof instances whatsoever.No Methods

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

No Type Checking!

• There is no type checking for subjects, predicates and objects!

• Examples:

• No mechanism prevents such a nonsense!

• Why?

- Semantic Web principle:

“Anyone can say anything about any topic”

(similar to classic web pages)

43

wd:Q5592 :date_of_birth "Nonsense" .

:person :person :person .

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Properties in RDFvs. Object-Orientation

• Properties comparable with associations between class instances

• But there are major differences:

44

Object-Orientation Semantic Web

Associations between instances arespecified in class definition

No need for specifyingrdf:type rdf:Property

Types of association ends arespecified

No need for specifying rdf:domainand rdf:range

Type violations result in errors(compile time or run-time)

No type checking (neither compiletime nor run-time)Instead: Reasoning results may benot as expected

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Serialization syntaxes for RDF

• Turtle / N3 (most convenient; used in this course):

• N-Triples

• RDF /XML

45

:person rdf:type rdfs:Class .

<http://h-da.de/fbi/artontology/person>

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

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

<rdf:Description

rdf:about=" http://h-da.de/fbi/artontology/person">

<rdf:type>http://www.w3.org/2000/01/rdf-schema#Class

</rdf:type>

</rdf:Description>

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201946

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Literature

• Dean Allemang, James Hendler:

Semantic Web for the Working Ontologist:

Effective Modeling in RDFS and OWL.

Morgan Kaufmann Publishers Inc., San Francisco

• Bernhard G. Humm: “Applied Artificial Intelligence

– An Engineering Approach”. LeanPub, 2016

https://leanpub.com/AAI ,

Chapter 2: Knowledge Representation

47

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201948

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Knowledge RepresentationServices Map

49

Knowledge Base

Kn

ow

led

ge

Re

sou

rce

s

Query Engine Reasoner

API

Kn

ow

led

ge

Ed

ito

r

Integrated Environment

Da

ta I

nte

gra

tio

n /

S

em

an

tic

En

rich

me

nt

Editing ontologies Storing ontologies Inferencing Integratingknowledgeresources

Off-the-shelfontologies

KB access via programming

language

Product bundle for many knowledgerepresentation services

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Knowledge RepresentationProduct Map

50

Knowledge Base

Kn

ow

led

ge

Re

sou

rce

s

Query Engine Reasoner

API

Kn

ow

led

ge

Ed

ito

r

Integrated Environment

Da

ta I

nte

gra

tio

n /

S

em

an

tic

En

rich

me

nt

Topbraid Suite, OntoStudio, …

Protégé, TopbraidComposer, …

Pellet, FaCT++, HermiT,..

Virtuoso, GraphDB, AllegroGraph, rdf4J, JENA, …

Dbpedia, YAGO, CYC, Wikidata, GND, …

Fuseki,..

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Product Map as a table

51

Agenda

1. Overview

2. RDF

3. ArtOntology

4. SPARQL

5. Discussion

6. Literature

7. Services map / product map

8. Quick check

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.201952

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Quick checkKnowledge Representation (1/3)

1. What is the purpose of knowledge representation?

2. Give examples for classes, individuals, relationships, and rules

3. What is an ontology?

4. Name different knowledge representation approaches

5. What does reasoning mean?

Give an example

7. What is a resource in RDF?

8. How are namespaces used?

9. What is an RDF triple?

10. How are classes declared in RDF(S)?

How are individuals (instances) assigned to classes?

(tbc…)

53

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Quick checkKnowledge Representation (2/3)

11. How is the result set of SPARQL queries structured?

12. How are path expressions used in SPARQL?

13. How to query a transitive closure in SPARQL?

14. What additional features does SPARQL offer?

15. Explain open world assumption and closed world assumption.

What are the differences?

What are implications of the assumption made?

16. What are differences between classes / instances in RDF(S) and in

object-orientation?

17. How are properties declared in RDF(S)?

How are properties used?

(tbc…)

54

Prof. Dr. Bernhard Humm, Darmstadt University of Applied Sciences. www.fbi.h-da.de/~b.humm. 30.04.2019

Quick checkKnowledge Representation (3/3)

18. Name the „Semantic Web principle“. What are the implications?

19. Which serialization syntaxes exist for RDF?

20. Name knowledge representation services

21. Name a few products that implement those services

55