CAiSE 2004

Adaptive Web-based Courseware Development Using Metadata

Standards and Ontologies

The AdaptWeb Project is a consortium of two Universities supported by Brazilian Research Council, CNPq and UFRGS.

The AdaptWeb Project is a consortium of two Universities supported by Brazilian Research Council, CNPq and UFRGS.

Lydia Silva MuñozFormer M.Sc. Student

José Palazzo M. de OliveiraProfessor at Federal University of RGS

Porto Alegre - Brazil

2/21

CAiSE 2004

Adaptive Courseware

! An Adaptive Courseware can:– Adapt the system behavior to the profile of the students

! An Adaptive Web based Courseware can also:– Interoperate using Web resources enabling the reuse of

educative material created in the system context or in the context of other applications

3/21

CAiSE 2004

Adaptation tools

! A model of the Student Profile to conveys the knowledge about the student A model of the knowledge about the educative content based on metadata descriptions to convey:– the meaning of each piece of content– the correct way to use it (e.g. minimum speed connection)– the possible ways to be assembled with others in order to

obtain more complex learning object (e.g. courses) based in simple ones (e.g. topics explanations or exercises)

4/21

CAiSE 2004

Interoperation tools

! A Standard Metadata Model envisaged to describe e-learning content in a common language. (e.g. The Learning Object Metadata Model - LOM)

! A binding to implement the Standard Model on the Web. (e.g. an RDF binding)

! If needed, the definition of an Application Profile of the Standard Metadata Model that makes the standard suitable to the particular community without loss of compatibility

5/21

CAiSE 2004

Implementing Adaptation

Knowledge Space Hyperspace

6/21

CAiSE 2004

The AdaptWeb Hyperspace

Discipline Structure

Discipline.dtd

20

Topic Structure

Topic.dtd

Topic.dtd

Topic.dtd

Topic Support

(Lectures,Exercices, Examples, etc.)

7/21

CAiSE 2004

Content Adaptation

Index

Concept 1 Concept 2 Concept 3

Concept 2.1 Concept 2.2

Concept 2.1.1

Index

Concept 1 Concept 2

Concept 1.1

Concept 2.2.1

Concept 2.1 Concept 2.2

Engineering Physics

8/21

CAiSE 2004

Index

Concept 1■ Exercises" Examples# Material C.

Concept 2■ Exercises" Examples# Material C.

Concept 1.1■ Exercises" Examples

Concept 2.1■ Exercises" Examples# Material C.

Concept 2.2

" Examples# Material C.

Concept 2.2.1

■ Exercises

# Material C.

Index

Concept 2

♦ Exercises$ Examples% Material C.

Concept 3

♦ Exercises

% Material C.

Concept 2.1

♦ Exercises$ Examples% Material

Concept 2.2

$ Examples% Material

Concept 2.1.1

♦ Exercises$ Examples% Material

Concept 1

♦ Exercises$ Examples% Material C.

Content Adaptation

9/21

CAiSE 2004

10/21

CAiSE 2004

AdaptWeb Architecture

DomainKnowledge

Adaptive Content Selection

ADAPTATION

Adaptive Presentation

Ontology

Authoring

RecommendedWeb

Resources

Web ResourcesRecommendation

STORAGE

Student KnowledgeOntology

Hyper-Space

xml html

Ontology Enrichment

Author

AutomaticMetadata

Generation

Log and Login

Student Monitoring Student

11/21

CAiSE 2004

General View

Knowledge Space Model

Cognitive Learning Style Identification

Cognitive Learning Style

Student Model

Settlement

Student Model

Knowledge Space Model

Design

Standard Application Profile

Define the Application

Profile of the Standard

Student Auto Assesment

Registered activityes of the

Student

HyperspaceInstances

Metadata Standard for Educative

Content

Automatic Metadata

GenerationMetadata Instances

Define the Learning Trajectory Workflow

Student Learning Trajectory Workflow

Hyperspace Model

Hyperspace Model

Design

Educative Content

Authoring

Metadata Augmentation

12/21

CAiSE 2004

Structuring Metadata

! A loosely coupled metadata repository supported by a Web Ontology

! Supports a powerful adaptation mechanism– Inference can be used to achieve adaptation (e.g. transitive

properties indicating prerequisite conditions can be automatically computed)

! Supports reusable learning objects– Metadata based on standard vocabularies maintained on

the Web as reference points for semantics enable the resolution of interoperation at the semantic level

13/21

CAiSE 2004

Content Knowledge Ontology

awo:Topic

awo:isAvailableTo

awo:isAvailableTo

awo:customizes

awo:hasPrerequisite

awo:isPartOf

awo:learningPath

awo:Course

awo:Discipline

awo:LearningObject

awo:Support

awo:Exerciselom-

edu:Difficulty

lom-edu:difficulty

awo:Example

awo:Complementary

dcterms:RFC1766

awo:Contributor awo:NetworkConnection

lom-tech:TechnologyRequirement

xsd:Stringlom-edu:

InteractivityType

lom-tech:requirement

awo:creator

dc:descriptionlom-tech:location

lom-meta:metadataScheme

dc:language

lom-edu:interactivityType

lom:Entity

vcard:FN

awo:e-Support

awo:isRecommendedBy

awo:isPartOf

awo:supportsTo

lom-cls:Taxonomy

awo:subject

lom-edu:LearningResourceType

lom-edu:learningResource

Type

awo:Keyword

awo:keyword

rdfs:subClassOf

14/21

CAiSE 2004

Properties Refinement

rdf:Property

dc:relation

rdf:type

dcterms:requires

rdf:subPropertyOf

awo:hasPrerequisite

rdf:subPropertyOf

awo:isPrerequisiteOf

daml:inverseOf

! Inverse and transitive properties are explicitly declared

! Any Web agent (e.g. an RDF agent) can understand the super-property dcterms:requires and to interpret the ontology relation awo:hasPrerequisite with the more general semantics of the known property

15/21

CAiSE 2004

Properties Refinement

rdf:Property

dc:source

dc:relation

dcterms:requires

awo:LearningPath

awo:supportsTo

dc:subjectdc:creator

awo:hasPrerequisite

dcterms:isVersionOf

awo:Customizes

dcterms:hasVersion

awo:isCustomizedBy

dcterms:isPartOf

awo:isPartOf

dcterms:hasPart

awo:hasPart

awo:isSupportedBydaml:inverseOf

awo:isPrerequisiteOf

daml:inverseOf

awo:prevLearningPath

awo:isAvailableTo

awo:hasAvailable

daml:inverseOf

dc:contributorlom-life:educationalValidator

awo:isRecommendedBy

awo:recommends

daml:inverseOf

awo:creatorawo:subject

rdf:subPropertyOf

rdf:type

awo:keyword

daml:inverseOf

daml:inverseOf

daml:inverseOf

16/21

CAiSE 2004

Automatic Metadata Generation

! Each time new educative content is created into the system context a wrapper, aware of the semantics of the XML tags of the Hyperspace files and the creation context (e.g. the mother language of the teacher to infer the language of the learning object), automatically generates RDF statements describing learning objects in the Knowledge Space, such as:– . . .

17/21

CAiSE 2004

Automatic Metadata Generation

! . . .• 1) Disciplines and Courses plus the information of which

courses customize which disciplines• 2) Topics explanations, exercises, examples and other kind of

complementary material plus the information of which material support which topic explanation and the other relations

! The correct sequence of topics into a discipline and theisPartOf relation among them are inferred from the XML sequencing and nested position of elements

18/21

CAiSE 2004

Automatic Metadata Generation

ALGEBRA

LINEAR ALGEBRA

ABSTRACT ALGEBRA

LINEAR DEPENDENCY

LINEAR TRANSFORMATIONS

LINEAR SYSTEMS

CANONICAL FORMS

LINEAR SYSTEMS DIRECT SOLUTION METHODS

LINEAR SYSTEMS ITERATIVE SOLUTION METHODS

GAUSS METHOD

GAUSS METHOD WITH PIVOTING

LU FACTORIZATION

CHOLESKY FACTORIZATION

JACOBI METHOD

Linear Sistems

Introduction Direct Methods

subject

subject subject

Gauss Method

subject

Knowledge Space

Content Knowledge Ontology Instance

Domain Taxonomy Instance

Numerical Methods

subject

partOf relation

Hyperspace

NumericalMethods.

xmlLinear

SystemsOfEquations.

xml

Introductionxml

DirectMethods.

xml GaussMethod.

xml

19/21

CAiSE 2004

Student Ontology

Student Ontology Class

Content Knowledge Ontology Class

st:Student

st:CognitiveLearning Style awo:Course

awo:Topicawo:NetworkConnection

st:hasLearningStyle

st:hasKnowledgeOn

xsd:String

st:locationLearningTrajectoryWF

xsd:Boolean

st:wantsTutorial

awo:hasNetworkConnection

st:hasLearningGoal

20/21

CAiSE 2004

Final remarks

! This paper describes an in progress project for adaptive content automatic computation of Web-based courses, according to selected programs and student’s profile

! A new application profile of the Standard LOM based in a RDF binding is defined and implemented in order to give the system the capacity of share learning objects across the Web

! A formal Web ontology supports metadata descriptions based in the constructed application profile. Such a ontology sets the stage to use inference in the computation of adaptability

21/21

CAiSE 2004

Final remarks

! A simple taxonomy models the learned domain. Links relating each learning object with elements in the taxonomy indicates to which topic in the domain the learning object is associated

! Authoring software is provided for syllabus generation, supported by XML standard files (i.e. The Hyperspace)

! A wrapper automatically generates the metadata instances (i.e. The Knowledge Space). The automation is based on the XML structure of the Hyperspace and the knowledge available about the context creation of the educative content

! The learner’s profile is modeled by a Web Ontology that support adaptation tasks