Content uploaded by Owen Conlan
Author content
All content in this area was uploaded by Owen Conlan
Content may be subject to copyright.
Reusing Adaptive Learning Resources
Dietrich Albert*, Cord Hockemeyer**, Owen Conlan***, Vincent Wade***
* Hiroshima University, Japan, Dep. of Learning Science <dalbert@hiroshima-u.ac.jp>
** University of Graz, Austria, Department of Psychology <chockemeyer@acm.org>
***Trinity College Dublin, Ireland, Knowledge and Data Engineering Group
<Owen.Conlan|Vincent.Wade@cs.tcd.ie>
Abstract
Metadata standards allowing the description, discovery, management and reuse of
learning objects are of focal interest in the educational domain. However, current
standards do not reflect recent developments in e-learning stressing the importance
of adaptability of learning resources to the learners' needs and preferences. Within
the EASEL project (URL: http://www.fdgroup.com/easel), our objective is to
extend and use current metadata standards to support the discovery of adaptive
content as well as its management and reuse. Thus, in this paper, we focus on
specifications describing the adaptivity of learning contents. A generic extension for
current metadata schemas is suggested that is independent of the pedagogical model
underlying the adaptivity.
Keywords: Application of Metadata, Reuse, Adaptive Contents
1. INTRODUCTION
Due to the high cost of producing high quality multimedia learning material, the reuse of learning objects has
gained much interest and importance recently. Starting with regional (e.g. ARIADNE, [1]) or business oriented
(e.g. AICC, [2]) schemas for learning object description, different metadata standards have been developed in
order to support exchange and reuse of the resources. Recently, standards or specifications with a broader scope
based on the aforementioned work have been developed, e.g. by the IEEE LTSC [3] and by IMS-Project [4].
However, all of these metadata specifications are oriented towards describing static content.
Research results from psychology and pedagogy as well as advances in the design of electronic learning material
have lead to the development of content that can be adapted to the individual learner (see [5,6] as examples for
adaptive systems based on one such psychological model). However, there are a vast variety of different
approaches with different objectives, objects, and underlying models for adaptivity. These advances have not yet
been represented in the development of educational metadata schemas. This paper describes a project that aims
to produce metadata schemas that aid the discovery and reuse of static and adaptive content. We propose a
generic extension to current metadata schemas that allows the description of arbitrary kinds and models of
adaptivity.
We will first give a short overview of the EASEL project, its architecture and its approaches to adaptivity.
Afterwards - we will introduce the proposed metadata schema extension for adaptive learning objects in two
steps, introduction of a plain extension illustrating the main idea for the generic description of adaptivity and
introduction of a more elaborated extension allowing for the description of adaptivity in a more structured and
detailed way.
2. REUSING LEARNING RESOURCES: THE EASEL PROJECT
The EASEL project (Educator Access to Services in the Electronic Landscape, [7]) can be described through its
three core objectives
1. Support lecturers in searching and selecting (locating) existing learning resources suitable for their
courses.
Albert, D., Hockemeyer, C., Conlan, O., & Wade, V. (2001). Reusing Adaptive Learning
Resources. In C. H. Lee & et al. (Eds.), Proceedings of the International Conference on
Computers in Education ICCE/SchoolNet2001 (Vol. 1, pp. 205–210).
2. Support lecturers in building new courses from existing materials through a Course Constructor Kit
(CCK).
3. Provide means for integrating adaptive material into newly built courses.
A simplified structure of the EASEL components is shown in Figure 1. The course author uses the CCK through
the WWW Course Constructor Client to look for appropriate learning material from local and remote
repositories via the search gateway. The selected material is then assembled and stored as a content package in
the Learning Management System (LMS) from where it can be accessed by the learners through a Learning
Environment (LE). In the case of adaptive resources, which are delivered via third party services (see below,
Section 2.2), the repositories contain only the metadata for these adaptive resources. This metadata is transferred
through the LMS to the LE. which then communicates with the adaptive service using the Content Interworking
API described below.
Metadata
Course
Constructor
Kit Metadata Based Search
W W W
Course Constructor
Client
Learning
Environment Adaptive
Third Party Service
Learning
Management
System
Adaptive Service
Learning Object
Repositories
Search Gateway
Content
Previews
Figure 1: EASEL Architecture
2.1. Interface for Adaptive Content
In most cases of adaptivity, the adaptive learning resource and LE communicate through a SCORM compliant
Content Interworking API [9]. This communication consists of two elements, the provision of learner
information for the adaptive learning resource by the LE, and the update of information on the learner's
performance and progress stored in the LE by the adaptive learning resource. The interface can be characterised
through three main phases:
1. Initialisation phase: The adaptive resource requests to initialise a new session.
2. Communication phase: The adaptive resource stores and retrieves variables and their values using a
common data model. The stored variables may be used by the same adaptive learning resource at a later
time or by a different resource. Any changes are only stored temporarily and locally to the current session
until changes are requested to be made permanently.
3. Finalisation phase: The adaptive resource requests to close the session after permanently storing all changed
variables and values. Such a permanent storage can also be performed within a session through a commit
command.
2.2. Different Approaches to Adaptivity within EASEL
Within EASEL, the different partners will take three different approaches to achieve adaptivity: pre-selection,
document-internal rules, and third party service. In the pre-selection approach, the concept of adaptivity is
understood in a broader sense than normally. Adaptivity here takes place during course construction, i.e. the
material is adapted to the specific course including the teacher's preferences and not to the individual learner
Opposed to the other two approaches explained below, the pre-selection is achieved offline and not during
runtime. Nevertheless, appropriate metadata describing the adaptive elements of this content is also required to
facilitate its effective discovery and reuse. With respect to the LE, this approach is the least demanding and most
portable one as the learning material is static by the time it is passed over from the CCK to the LMS and LE.
In the document-internal rules approach, the learning resources contain rules specifying the adaptivity. These
rules are then interpreted utilising the Content Interworking API to persistently store variables across multiple
pages (see above for an overview on this API). A similar approach is currently investigated within the CLEO
project [8]. This approach is the most demanding and the least portable one with respect to the LE as it must
provide the respective functionality.
In the third party service approach (TPS), the adaptive content is provided through an external service. The CCK
only stores a reference to the external service as part of the course in the LMS and LE. When a learner reaches
this adaptive part of a course, a connection to the TPS is launched. The TPS then retrieves information about the
learner and his knowledge, preferences, and progress from the LE, using the Content Interworking API, and
subsequently adapts the learning material according to these information.
3. A GENERIC METADATA SCHEMA EXTENSION FOR ADAPTIVE
CONTENTS
One major goal of the EASEL project is to extend educational metadata standards (e.g. IMS Metadata) such that
they can describe adaptive features of electronic learning resources. There exist a number of standards and
specifications issued by different communities and institutions, however, due to a large overlap in the activities
of standard developing bodies and working groups, these different standards are closely related to each other.
We decided to take the IMS Learning Resource Metadata v1.1 specification [10] as a basis for our work. One
reason was the fact that this is a specification for which an official XML encoding specification is also available.
Another reason being that IMS has efficient procedures for the development and update of its specifications, i.e.
a decision on proposed extensions is made quite quickly. Due to the overlap between the different existing
standards it should be easy to adopt the proposed extensions to the other standards.
3.1. A Basic Generic Adaptivity Element
We propose a new, optional element adaptivity within the education block of IMS Metadata (a first sketch of this
proposal has been given by Conlan et al., [11]). Figure 2 shows the structure of this new element. It contains an
arbitrary number of elements adaptivitytype each of which describes one type or aspect of adaptivity available
for this learning resource. The adaptivitytype element itself has two attributes (a mandatory name and an optional
ref), and a langstring content. The name denotes the type or aspect of adaptivity for which information is
provided while the langstring content contains the information itself. The second, optional attribute ref can be
used to specify a URI where the vocabulary used in the langstring content is defined. The possible values for the
name will be partially restricted by a best practice list.
Below, an adaptivity element for an imaginary content is shown containing several different adaptivitytype
entries. The competencies types elements in this example show that a hierarchical structure is possible. This
demands that the same vocabulary should be used for the langstring content of all competencies.XXX entries.
education
...
adaptivity?
adaptivitytype*
name=<langstring>
ref=<URI>?
langstring
Figure 2: The proposed generic adaptivity metadata element
<adaptivity>
<adaptivitytype name="specialneed">
Visual Disability
</adaptivitytype>
<adaptivitytype name="competencies.taught" ref="some-sameURI">
RDBMS Management
</adaptivitytype>
<adaptivitytype name="competencies.required" ref="some-sameURI">
Database concepts
</adaptivitytype>
<adaptivitytype name="learningstyle" ref="some-otherURI">
auditive
</adaptivitytype>
</adaptivity>
This example denotes a learning resource that offers four different types of adaptivity. The specialneeds entry
declares the usability of this learning resource for blind learners. The competencies.taught entry says that the
learner acquires the competency to manage a relational database management system learning with this resource,
i.e. the entry specifies the objectives of the learning object. A learning object may also have several possible
objectives from which the learner (or learning environment) may choose. The content of this competency
RDBMS management is described in more detail at the referred resource. The next entry specifies that the learner
should already have acquired the competency Database concepts before processing the current resource because
that knowledge is needed for understanding and successful processing. Such information may be used for
navigation support, i.e. for an adaptive sequencing of learning objects. The exact meaning of this competency is
defined in the same referred resource as the taught competency. This is important in order to ensure that taught
and required competencies can be related to each other (see the next section on vocabularies). The final entry
learningstyle claims that the resource is useful for learners preferring an oral presentation of contents.
3.2. A Structured Adaptivity Element
The generic element laid down in the previous section provides, in principle, a means for describing many
different approaches for adaptivity. However, for some approaches it would be advantageous to be able to
structure the information specified within a certain adaptivitytype entry. Regarding the specification of
competencies required to be able to understand a certain document, for example, should cover two aspects at the
same time. There may be several competencies necessary to be able to understand the document, and there may
be different approaches possible to understand this document resulting in several alternative sets of prerequisites.
As a consequence, the need to structure the data provided within the adaptivitytype entry arises. Figure 3
sketches the proposed extended adaptivitytype entry covering such structured data. This extension consists of
two parts addressing different reasons for possible problems in the interpretations of multiple langstring fields
within one adaptivitytype entry: A candidate block may contain several langstrings specifying the same value in
Figure 3: A structured generic adaptivity metadata element
education
...
adaptivity?
adaptivitytype*
name=<langstring>
ref=<URI>?
set?
type=<type-of-set>
set *
set*
candidate*
langstring*
different languages. A set then can contain several candidates with different meanings. The optional type
parameter of the set entry specifies how the candidates should be connected. Example values for this type are all
(i.e. a conjunction) or at-least-one (i.e. a disjunction). On top level, only one set is allowed in order to avoid
ambiguities about the combination of sets at this level. In principle, sets may be nested arbitrarily; however,
normally two set levels should be sufficient (any and-or structure can be represented by a disjunction of
conjunctions or vice versa) in order to reduce complexity of the provided data.
This structured approach is illustrated in the following imaginary example.
<adaptivitytype name"competencies.required">
<set type="at-least-one">
<set type"all">
<candidate>
<langstring lang="en">competence-A</langstring>
<langstring lang="de">Kompetenz-A</langstring>
</candidate>
<candidate> ... </candidate> ...
</set>
<set type="l"> ... </set> ...
</set>
</adaptivitytype>
On top level, we have a set of type at-least-one, i.e. in order to understand the current document, the learner
should master at least one of the following sets of competencies. Both second-level sets are of type all, i.e. the
learner should master all competencies within one of the sets. In the first candidate, we then also have an
example for specifying information in different languages. The competence A is specified in English as such,
while the German specifications names it as Kompetenz A.
3.2. Best Practice Lists and Vocabularies
The generality of the proposed extension has the disadvantage that users have to identify domain specific terms
in a unique but machine-recognizable way. With respect to the name attribute of the adaptivitytype entry, this
can be realized through a best practice list given that a simple but, nevertheless, moderated way for extending
this list is provided. Regarding the langstring content, this is more difficult because that will often depend on the
vocabulary of the knowledge domain under consideration, e.g. the competencies entries in the example above.
While there exist attempts to build a general ontology (e.g. in the IEEE SUO Group, [12]), these have not yet
shown satisfactory results that could be applied for this purpose. We therefore propose to use the ref attribute to
specify the vocabulary used in describing the learning object. Nonetheless, it is important that widely accepted
standard vocabularies for the different fields of knowledge are used because only standard vocabularies can
ensure the interoperability and information exchange within learning objects' metadata. Current experiences
within the EASEL project show that classification systems as used in libraries like UDC [13] are too coarse for
this objective.
4. CONCLUSION
In the EASEL project, a toolbox for constructing new courses from existing - static as well as adaptive - learning
objects is developed. The EASEL learning management system is based on metadata created according to
several existing educational metadata standards, IEEE LOM, IMS Metadata, and Dublin Core Education
standard recommendation.
We have proposed an extension for the IMS Metadata schema to accommodate a generic adaptivity element.
Based on a first, simple concept, practical requirements have led to a more elaborate approach which allows the
information specifying the adaptivity of a learning resource to be structured.
Currently, these proposals are realized in a trial implementation in order to prove their usability. Nevertheless,
there are still open issues, e.g. the controlled vocabulary problem that has not yet been satisfactorily solved.
Acknowledgments
This research is partially funded by the European Commission under the auspices of the EASEL (Educator
Access to Services in the Electronic Landscape, [7]) project. The EASEL projects goal is to explore technologies
which can be brought together to offer course constructors an environment in which they can readily combine
existing learning objects to create new online educational offerings. Current proprietary Adaptive Hypermedia
Services tend to restrict this kind of integration. As part of EASEL the research conducted will be used to
integrate Adaptive Hypermedia Systems into Learning Environments which are based on current WWW
educational standards.
Dietrich Albert is professor at the Department of Psychology, University of Graz, Austria. Currently, he is at
Hiroshima University, Japan, as a visiting professor.
The authors wish to thank Aida Slavic and Roberto Trabucchi for their comments on an earlier draft of this
paper.
REFERENCES
[1] Alliance of Remote Instructional Authoring and Distribution Networks for Europe (ARIADNE). EC FP4
project. URL: http://ariadne.unil.ch/.
[2] Aviation Industry CBT Committee (AICC). URL: http://www.aicc.org/.
[3] IEEE Learning Technology Standards Committee (LTSC). URL: http://ltsc.ieee.org/.
[4] IMS Global Learning Consortium. URL: http://www.imsproject.org/.
[5] ALEKS. URL: http://www.aleks.com/.
[6] Hockemeyer, C., Held, T., and Albert, D. (1998). RATH - a relational adaptive tutoring hypertext WWW-
environment based on knowledge space theory. In Christer Alvegård, editor, Computer Aided Learning and
Instruction in Science and Engineering, Proceedings of the CALISCE'98, pp. 417-423, Göteborg, Sweden:
Chalmers University of Technology.
[7] Educators Access to Services in the Electronic Landscape (EASEL). EC IST project 10051. URL:
http://www.fdgroup.com/easel/.
[8] Custumized Learning Experiences Online (CLEO) Lab. URL: http://www.cleolab.org/.
[9] Sharable Content Object Reference Model (SCORM), version 1.1. Advanced Distributed Learning, January
2001. URL: http://www.adlnet.org/.
[10] IMS Learning Resource Meta-data Specification, version 1.1, June 2000. URL:
http://www.imsproject.org/metadata/index.html.
[11] Conlan, O., Hockemeyer, C., Lefrere, P., Wade, V., & Albert, D. (2001). Extending educational metadata
schemas to describe adaptive learning resources. In Hugh Davies, Yellowlees Douglas, & David G. Durand,
editors, Hypertext 01: Proceedings of the twelfth ACM Conference on Hypertext and Hypermedia, pp. 161-162,
New York: Association of Computing Machinery (ACM).
[12] IEEE Standard Upper Ontology (SUO) Working Group. URL: http://suo.ieee.org/.
[13] Universal Decimal Classification (UDC), pocket edition. British Standards Institution, 1999.
