CONTENT PACKAGE ADAPTATION: A WEB SERVICES
APPROACH
Ricardo Fraser, Permanand Mohan
Department of Mathematics and Computer Science
The University of The West Indies
St. Augustine, Trinidad and Tobago
Keywords: e-Learning, Web Services, Adaptive Hypermedia, Content Package, Learning Services, Learning Object,
Personalisation, Accessibility
Abstract: The IMS Content Packaging Specification is a format that facilitates the deployment of discrete units of
learning resources based on an XML structure called a manifest. The contents and structure of a content
package are determined at design time when it is created. Since the package has been authored for use in a
particular instructional setting, re-purposing the content package to meet the demands of a different
instructional setting is difficult. Although there have been attempts to improve the flexibility of the package
such as using IMS Simple Sequencing, the adaptation provided is still inadequate. In this paper we argue
that Web Services can be used to facilitate the dynamic adaptation of a content package so that it can be
reused in diverse instructional scenarios and accessed by additional learners who otherwise would not be
able to utilize it. We present a framework for adaptation based on web services and identify a representative
set of web services that could be used for content package adaptation. We then discuss in detail the Media
Integration and Translation Services for Accessibility (MITSA), a category of web services designed to
promote media accessibility of a content package. Finally, we conclude by highlighting the benefits of the
web services approach for content package adaptation.
1 INTRODUCTION
An IMS content package (CP) (IMS Content
Packaging Information Model, 2003) is a format that
can be used for storing learning objects. A learning
object is commonly understood to be an independent
and self-standing unit of learning content that is
predisposed to reuse in multiple instructional
contexts (Polsani, 2003). A content package contains
a set of digital learning resources as well as a
manifest that describes how the learning resources
should be organized to form a unit of instruction. If
learning objects are stored in the form of an IMS
content package, the content package essentially
combines learning content into a discrete unit of
learning that could potentially be reused in different
instructional platforms and scenarios.
One approach for reusing learning objects stored
as co
ntent packages is to disaggregate and assemble
the learning resources contained in the package to
form new units of instruction. However a more
recent approach by Fraser and Mohan (2004)
suggests that a content package can be imbued with
knowledge about web services that can facilitate
dynamic adaptation when used in different
instructional scenarios. In this paper, we argue that a
web services approach can be used to provide a
more sophisticated form of content package
adaptation. This in turn promotes a deeper kind of
reusability of learning resources than what is
possible using the ‘pure’ learning objects approach.
We present a viable framework for content package
adaptation and identify a representative set of web
services that could be used to facilitate content
package adaptation. Additionally, we briefly review
and discuss other service-oriented frameworks for
adaptation in learning systems. We also present the
Media Integration and Translation Services for
Accessibility (MITSA), a category of web services
designed to promote media accessibility of content
packages.
238
Fraser R. and Mohan P. (2005).
CONTENT PACKAGE ADAPTATION: A WEB SERVICES APPROACH.
In Proceedings of the Seventh International Conference on Enterprise Information Systems, pages 238-243
DOI: 10.5220/0002529302380243
Copyright
c
SciTePress
2 CONTENT PACKAGING
The IMS content packaging specification was
designed to address interoperability issues with
regard to the import and export of aggregate and
disaggregate packages of content between learning
management systems (LMS). It is a standardised
format for specifying the assembly of learning
resources into an interoperable unit at different
levels of granularity. A content package consists of a
single top-level
manifest file together with the
physical files it specifies. It is shipped via a single
digital resource called a package interchange file
(PIF), which could be in some appropriate archive
format such as zip, cab or jar. The basic structure of
a content package is shown in Figure 1. The function
of the manifest is to specify the structure of the
content in the package by means of an organizations
section (currently, only hierarchical structures are
permitted). The resources section of the package
identifies the physical files that comprise the content
package. It may also identify URLs, which are links
to external resources. Items in the content structure
are linked to the resources that are required for
rendering the content on a learner’s web browser.
The manifest is serialized in XML and optionally
describes a set of simple sequencing rules and a
learning design. Additionally, the content package
may contain metadata that describes its
characteristics. The IEEE Learning Object Metadata
(LOM) standard can be used for this purpose (IEEE
LTSC). Metadata instances can be used to describe
the entire content package (i.e., learning object) or
individual resources and items within the package. A
resource may also refer to a sub-manifest. Thus, a
manifest may consist of other sub-manifests, each of
which is a content package in its own right. This
enables the aggregation of learning resources into
coarser-grained structures of learning.
In its present form, the IEEE metadata standard
does not contain enough information to allow re-
purposing of a content package in a new
instructional scenario (Farance, 2003). When a
content package is assembled, it is intended for use
in a particular context and instructional scenario. A
unit of instruction refers to a content package with
an appropriate instructional design. Thus, a learning
object is really a unit of instruction at different levels
of granularity. When learning objects are dis-
aggregated and assembled into new content
packages, the instructional design that applied to the
original content package is no longer applicable. It is
important to note that some researchers (Koper,
2001) do not consider a learning object that does not
have an associated relevant learning design to be a
unit of instruction. The specification does not
address this limitation.
Another limitation is that the content package
specification does not facilitate the dynamic
adaptation that responds to the changing needs of
learners. Since the delivery structure has been
designed and shipped with the package (i.e., the
items of content and the physical media associated
with those items have already been determined), it is
not possible for the LMS to adapt the content
package to new learner requirements. This is also
true when an item of content references an external
resource since the same resource is shipped to every
learner whether or not a server has prepared it
dynamically. This is the classic, traditional client-
sever paradigm of server-side applications. The
problem is that the resource is independent and
largely irrelevant of the new instructional scenario
since the request for the resource on a web server
has been hard-coded in the CP and cannot change.
Thus, the assumption that a CP is a reusable
resource is misleading since it cannot be re-purposed
in a timely manner. The main goal of the paper is to
address the latter limitation and show how more
sophisticated forms of reusability are permitted
using a web services approach.
PACKAGE
MANIFEST
(an XML file
that consists of
metadata and a
catalog of
component
resources)
PHYSICAL
FILES
(The actual
digital contents
and other files
of the package)
3 ADAPTIVE LEARNING
SYSTEMS
An adaptive learning system (ALS) is one that
facilitates a learning environment that meets the
goals and objectives of the learner by presenting
pedagogically sound learning material as the needs
of the learner change. Since the learning materials
on the WWW commonly take different media forms,
ALS are sometimes called Adaptive Hypermedia
Systems (AHS). ALS is based on a move away
from the “one size fits all” thinking as regards the
distribution of learning material (Brusilovsky, 1996).
The navigational freedom in conventional
hypermedia applications leads to comprehension and
Figure 1: IMS content package – basic structure
CONTENT PACKAGE ADAPTATION: A WEB SERVICES APPROACH
239
orientation problems (Brusilovsky, 1996). Adaptive
hypermedia attempts to overcome these problems by
adapting the presentation of information and the
navigation structure, based on a user model
(Brusilovsky, 1996). This guidance and direction
essentially channels and fosters the cognitive links
that learners seek to create in order to truly
understand the concepts the media aims to teach.
The most popular techniques used in ALS can be
classified into two main types (De Bra, Brusilovsky
and Houben, 1999). These are:
Adaptive presentation - system determines
what the user currently experiences
Adaptive navigation support – system
decides where the user goes next
The user model, which importantly measures the
student’s grasp or knowledge state of a concept,
drives the adaptation.
For adaptive navigation support, the user model
is applied to classify learning items into several
groups according to the user's current knowledge
state and interests or goals. The ALS manipulates
the link structure to guide users towards interesting,
relevant information (Brusilovsky, 1996). A classic
example of this strategy is in a recently proposed
architecture that identifies and groups learning
elements that share common concepts into candidate
groups. The user model then motivates the system to
guide the user to learning items in a particular
candidate group (Dagger, Conlan and Wade, 2003).
Other techniques under this category include link
hiding, link annotation, link disabling, and link
removal (Brusilovsky, 1996).
Adaptive presentation adapts the presentation of
the current page according to the user model
particularly the knowledge state. It is typically used
to provide “scaffolding” for the learner. Scaffolding
is a term used to provide additional support to
learning beyond the mainstream delivery. This may
include providing prerequisite, additional, or
comparative explanations for “hard to understand”
concepts (Brusilovsky and Maybury, 2002).
Techniques used to provide adaptive presentation
include:
STUDENT
Pedagogic
Agents
(Decisions
for CP
adaptation)
Web
Services
(Execution of
CP)
ADAPTIVITY
SERVERS
U
S
E
R
M
O
D
E
L
E
V
E
N
T
S
LMS
ADA
PT
ED
CP
S
E
R
V
I
C
E
S
-
A
N
N
O
T
A
T
E
D
C
P
(ADAPTED)
RENDERED CP
Conditional inclusion of learning fragments
based upon the user’s knowledge state (De
Bra, Aerts, Berden and de Lange, 2003)
Stretch text: system initiated stretching or
shrinking of learning fragments and
subsequently controlled by user (De Bra,
Brusilovsky and Houben, 1999)
Providing alternative explanations
(Vassileva and Deters, 1998)
Reordering of presented items on a page
(De Bra, Brusilovsky and Houben, 1999)
Figure 2: Content package adaptation via web services
4 WEB SERVICES ADAPTATION
4.1 Overview of Web Services
A web service is a service provided on the Web by a
server computer that may be invoked by several
clients in a distributed framework. In order to
communicate with a web server that provides a
service, a client uses SOAP (W3C, 2004) - a
packaged, XML-formatted method call to a service
on a particular server. The web server provides a
catalogue of available methods from which the client
may choose to access a particular service. This
catalogue is in XML format and conforms to a
platform independent Web Services Description
Language (WSDL) (W3C, 2003), which is an API
for the service invocation. It indicates a parameter
list and return type of each method belonging to a
particular web service. After executing the client’s
request encoded in a SOAP envelope, the web
service replies to the client using a SOAP message
of its own. The client then unravels the SOAP
package to retrieve the response. In this way, web
services provide a run time environment for
executing methods over an HTTP or HTTPS wire
(i.e., over the Web).
ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING
240
4.2 Architecture
We propose that web services can act as a runtime
environment for the execution of services identified
in a CP. CP services are standardised XML
formatted tags that identify the resource to be
adapted, a description of the specific adaptation to
be carried out and the actual endpoint address of the
web service that carries out the adaptation. These
service tags are invisible to the LMS so that they do
not intrude on the traditional model of interaction
that exists between the LMS and the CP. In the
classic model of interaction, the LMS receives a
content package from a digital repository and simply
renders it to the learner. See Figure 2.
Software agents use the information recorded in
a CP to actually invoke the web services specified
by the endpoint address. These agents are aware of
the services specified in the CP and continually
monitor the user model (student profile) for
noteworthy changes. There are also agents that
create, manage and update the user model based on
the learner’s interaction with the system.
Information in the user model may include the user’s
learner preferences and style, competencies,
objectives, learning history and browser and client
capabilities. Agents that determine user model
changes that warrant adaptation invoke the web
service endpoint address in the CP - called a CP web
service (CPWS). When the web service has carried
out the adaptation of the CP (or part of it), the result
is returned to the adaptation agent. This agent then
forwards the adapted CP to the LMS so that it can
render the newly adapted learning resource. Since
the user model drives the adaptation, the new change
improves the user’s learning experience and leads to
better learning. Thus CP services, via specified web
services, dynamically re-purposes reusable online
learning resources by adapting the CP to learner
needs (Fraser and Mohan, 2004). The capability of
the web client to correctly display any adapted
content also affects the agent’s decision-making
process for adaptation. In the main, this web services
approach provides dynamic (on the fly) adaptation.
Particularly, it provides a powerful mechanism for
adaptive presentation. It is thus an effective service-
oriented approach to adaptive e-learning.
It may be necessary to procure learning resources
based on the type of adaptation they are capable of
undergoing. Thus, adaptive capability forms part of
metadata search criteria and digital repositories can
be queried for specific service types. We will now
suggest some candidate web services for e-learning
and look closer at the MITSA services.
5 WEB SERVICES AND
IMPLEMENTATION
5.1 Candidate Web Services
To support our argument that the web services
approach is a viable means of providing dynamic
and meaningful content package adaptation, we
present a non-exhaustive list of representative
classes of basic web services for e-learning. These
are listed in Table 1.
5.2 Overview of MITSA
There is a need to provide quality educational
content to the differently able. However in most
cases, the educational material is not aimed at
persons who possess hearing and visual impairment
or some other form of impairment. Both the W3C
Accessibility Initiative Policies and Web Content
Accessibility Guidelines (W3C WAI, 2004) strongly
encourages that content providers on the WWW take
this target audience into account in the design of
Table 1: A representative set of web services for content package adaptation
Candidate Set of Web Services
Service class Functional Description Benefits
Language Translation
Services (LTS)
Translate the language of the resources
and any associated metadata
Extend the reach of learning resource
to speakers of diverse languages
thereby increasing reusability
Media Integration and
Translation Services for
Accessibility (MITSA)
Transform media type of learning
resource from one to the next
Makes content accessible to a new
learner type and increases the didactic
effectiveness of resource
Dynamic Metadata
Adaptation and Generation
for Interoperability (DMAGI)
Translate one metadata profile to
another, adapt metadata profiles to a
change in resource, generate new
profile from resource
Improve interoperability between
learning systems, promote better
resource discovery and reusability
CONTENT PACKAGE ADAPTATION: A WEB SERVICES APPROACH
241
their sites and products. The policy also generally
defines requirements for educational content
providers.
MITSA aims to fulfil the need to reach a broader
range of individuals including those who could not
otherwise access quality educational content because
of the alienation imposed upon them by traditional
forms of text and other multimedia. It facilitates
translation from one multimedia format to another
and also reintegrates disparate pieces of processed
multimedia so that the resulting product is
meaningful and pedagogically viable. MITSA is
intended to widen the reach of reusable learning
content by distributing educational content equitably
thereby bridging the accessibility barriers.
Consider the following use case of MITSA. This
use case is for ‘normal’ learners who wish to get a
text to voice translation of a learning resource. First,
the learner clicks on an item from the content
package (usually generated as a hierarchical frame
display by the LMS). The monitoring agent senses
the request and on checking the user-modelling
agent, determines that the learner has new
preferences for both textual and voice formats of a
learning resource or certain competency levels
demand both. An agent then retrieves the
appropriate learning resource from the LMS. The
adaptation agent then examines the content package
and finds the web service identified for translation.
This web service is invoked with the learning
resource as a parameter. The web service returns its
results in the form of an audio file, which is passed
to the LMS for delivery. The LMS then streams the
audio file to the learner’s browser and
simultaneously displays the textual content.
5.3 MITSA Implementation
The MITSA implementation is based on the
architecture presented in Section 4. The
implementation involves the inclusion of a services
element in the general element of the metadata
section of a CP, its individual items or resources.
MITSA and other candidate services implement the
CP services alluded to by the architecture in a way
that extends the CP specification. Examples of
services that reside under the umbrella of MITSA
include text to voice services such as that described
in the use case and text to Morse code services.
Consider the sample excerpt of content package
metadata that elaborates one MITSA service element
below.
<cpws:services>
<service class=”cpws:mitsa”>
<parameters>
<parameter type=”resource”>
<id>resource1</id>
<mime-type>text/plain</mime-type>
<encoding>ASCII</encoding>
</parameter>
</parameters>
<loc>http://sage.uwi.tt/cpbehave/audio</loc>
<method>toAudio</method>
<description>text to audio</description>
<return>
<mime-type>wav</mime-type>
<encoding>base64</encoding>
</return>
</service>
</cpws:services>
The agent gathers the necessary information for
adaptation by parsing the cpws:services element. In
the above example, the agent recognises the service
as belonging to the MITSA class. The web service
will take one parameter, which is resource1 and is
plain text. The name of the web service that will
execute the action in the description element is
toAudio. It will return a wav file but it will be in
base64 format. This information is necessary to
inform the agent how to store and deploy the media
to the LMS. Many service elements are possible
each belonging to its own class. This is a simple yet
very powerful example of adaptive presentation
using web services. We will now discuss and
compare other service-oriented approaches to
adaptive e-learning.
6 DISCUSSION
Other researchers are increasingly recognising the
value of learning services in adaptive e-learning. For
example, a system (Dolog, Henze, Nejdl and
Sinteky, 2004) provides a framework for
personalized e-learning using web services and
describes an architecture for personalised learning
support in a P2P, distributed e-learning environment.
They identify two sets of services. These are
personalization services and supporting services.
Web services are used as a means of carrying out
personalization functionalities for personalized
learning support. However they are not used to
directly adapt a learner resource. That is, their
approach does not use services in a way that extends
the reach of the CP (learning resource) for
accessibility by increasing its plasticity for different
learning types.
A major benefit of the web services approach is
the quality of adaptation that may be performed by
the most powerful and sophisticated software that
underlie the web services. For example, for the
MITSA, Java Web Services are used and the text to
voice conversions are done by software that uses the
Java Speech API (JSAPI). There are several third
party software firms who specialise in the creation
ICEIS 2005 - SOFTWARE AGENTS AND INTERNET COMPUTING
242
and execution of particular services. This
specialisation of labour leads to greater productivity
as is well known in the field of economics. As the
learning services industry becomes lucrative other
firms will be attracted to participate. As more
participate, competition and market forces will cause
service provision and execution costs to learning
systems to become cheaper. The quality of services
increases further as firms attempt to remain
competitive in their service provision and execution
portfolios for education. In sum, the web services
approach promises affordable, personalised and
adaptive quality learning.
7 CONCLUSION
The web services approach has been presented in
this paper as a viable means of extending the reach
of a rigidly and statically designed content package.
Much effort has been spent in specifying standards
for packaging learning objects and describing them
using metadata. Our approach integrates well with
these existing standards. It simply extends the
content package with information related to services
that can lead to dynamic adaptation in diverse
instructional scenarios. In doing so, it takes
reusability and adaptability to a new level, allowing
service providers on the Web to augment the
potential of learning objects in many different ways,
media translation and formatting being one of them.
We feel that providing services that can enrich
content in a more controlled environment is
potentially a more viable means of reusability than
the pure learning objects approach that has been
promoted over the past few years. Of course, the
development of tools to annotate these services in
the learning object will encourage adoption of the
approach.
REFERENCES
Brusilovsky, P., 1996. Methods and techniques of adaptive
hypermedia. User Modeling and User-Adapted
Interaction, 6 (2-3), pp. 87-129
Brusilovsky, P. and Maybury, M. T., 2002. From adaptive
hypermedia to adaptive Web. In P. Brusilovsky and
M. T. Maybury (eds.), Communications of the ACM
45 (5), Special Issue on the Adaptive Web, 31-33.
Dagger, D., Conlan, O., Wade, V., 2003. An Architecture
for Candidacy in Adaptive eLearning Systems to
Facilitate the Reuse of Learning Resources. In E-
Learn 2003, World Conference on E-Learning in
Corporate, Government, Healthcare and Higher
Education, November 7-11, 2003, Phoenix, Arizona,
49-56.
De Bra, P., Aerts, A., Berden, B. and de Lange, B., 2003.
Escape from the Tyranny of the Textbook: Adaptive
Object Inclusion in AHA! In proceedings of the
AACE ELearn 2003 Conference, Phoenix, Arizona,
November 2003, pp. 65-71.
De Bra, P., Brusilovsky, P. and Houben, G.-J., 1999.
Adaptive Hypermedia: From Systems to Framework.
In ACM Computing Surveys, 31 (4es) December
1999.
Dolog, P., Henze, H., Nejdl, W., & Sinteky, M., 2004.
Personalization in Distributed eLearning
Environments. In WWWW2004, World Wide Web
2004, May 17–22, 2004, New York, New York, USA.
ACM 1581139128/04/0005.
Farance, F., 2003. IEEE LOM Standard Not Yet Ready for
'Prime Time'. In Learning Technology, 5(1), January
2003.
Fraser, R. and Mohan, P., 2004. Using Web Services for
Dynamically Re-purposing Reusable Online Learning
Resources. In ICALT ‘04, 4
th
IEEE International
Conference for Advanced Learning Technologies,
Joensuu, Finland, Aug 30-Sep 02, 2004, 595-599.
IEEE Learning Technology Standards Committee, IEEE
LTSC, http://grouper.ieee.org/groups/ltsc/index.html.
IMS Content Packaging Information Model, Retrieved
November 8
th
2004, from the World Wide Web:
http://www.imsglobal.org/content/packaging/.
Koper, R., 2001. Modelling Units of Study from a
Pedagogical Perspective: the pedagogical meta-model
behind EML. Retrieved November 8
th
2004, from the
WWW: http://eml.ou.nl/introduction/articles.htm.
Polsani, P., 2003. Use and Abuse of Reusable Learning
Objects. Journal of Digital Information, 3(4).
Vassileva, J. and Deters, R., 1998. Dynamic Courseware
Generation on the WWW. In British Journal of
Educational Technologies (1998), 29 (1), 5-14.
W3C, World Wide Web Consortium, http://www.w3c.org.
W3C WAI, 2004. W3C Web Accessibility Initiative,
http://www.w3.org/WAI/.
CONTENT PACKAGE ADAPTATION: A WEB SERVICES APPROACH
243