Content uploaded by Paolo Maresca
Author content
All content in this area was uploaded by Paolo Maresca on Aug 26, 2014
Content may be subject to copyright.
Experience with the GESTALT
On-Line Learning Support System
*
Almerindo GrazianoIr2, Paolo Maresca', Stefan0 Russo'
'
Dipartimento di Informatica
e
Sistemistica, Universiti di Napoli Federico 11,
Via Claudio
2
1,80125
Napoli, Italy,
{
paomares, Stefano.Russo1 @unina.it
School of Computing and Management Sciences, Sheffield Hallam University,
Sheffield,
UK,
a.graziano@shu.ac.uk
Abstract
With the spread of the Internet and the Web, higher
education institutions are increasingly interested
in
exploiting modern standard information and
communication technologies to provide learners and
trainees with advanced services for tele-education. Within
the European Union 4th Framework Programme, the
project GESTALT has defined and implemented a
flexible, component-based architecture aimed at
supporting on-line discovery and delivery
of
multimedia
educational resources on the web. In this paper, we
describe the on-going experience with the GESTALT
system at Universiv of Naples, and report first results
of
on-field trials during the first academic semester.
1.
Introduction
The on-going revolution leading
to
the so-called
information society, driven by advances of computer and
communication technologies, is also affecting the fields
of higher education and vocational training. A number of
forces push educational institutions like universities
towards a shift in their business process model, from
institution-centred
(ie, built around the lecturer and the
classroom)
to
learner-centred
(built around the student,
who is not anymore a merely passive learner). Examples
of such forces are:
the increasing need for
permanent education and
training,
involving mature students, often employed,
willing to gain further skills and qualifications;
the need for
distance learning,
to
reach learners that
are remote, or disabled,
or
part-time students, and to
allow asynchronous interaction between the tutor and
the learners;
the availability of a huge amount of digital
information at hand through the web.
The Internet, the
WWW
and related technologies play
a fundamental role in this emerging scenario.
So
called
on-line learning
initiatives, fostered by individual
or
federated institutions, are established in many countries,
providing distance education opportunities.
We believe that, from a technical point
of
view,
crucial to the successful use of information technologies
for distance education is the adoption of open standards
and
tools,
and that these should interoperate
to
support a
number of activities (ranging from digital courseware
authoring to retrieval and delivery) accomplished by a
variety of players involved in the educational process,
including lecturers, tutors, students and staff. In this
paper, we describe the on-going experience with the
GESTALT learning
support
system at University
of
Naples. This work is part
of
the ACTS Project GESTALT
(Getting Educational Systems Talking Across Leading-
edge Technologies) of the EU
4*
Framework. Programme
[l].,
Many pedagogical issues and problems are also
clearly related to the use of information technologies for
education. These are beyond the scope of the project and
of this work in particular.
The paper is structured as follows. Next section
presents the GESTALT component-based architecture
and its functionalities, aimed at supporting on-line
discovery and delivery of multimedia educational
resources on the web. Section
3
briefly describes the
GESTALT metadata model underlying the discovery
services, and its relationships
to
standards. Section
4
describes the GESTALT testbed and trials at the
University of Naples, while in section
5
we comment on
the experience made; finally, section
6
gives some
concluding remarks.
'
This work
has
been supported by EU under Project AC367 GESTALT (Getting Educational Systems Talking Across Leading-edge Technologies)
of
the
ACTS programme (Advanced Communication Technologies and Services),
IV
Framework Programme.
86
1089-6503/00 $10.00
0
2000
IEEE
2.
The GESTALT architecture
The GESTALT functional model supports a number of
educational actions, roles, and relationships in a common
educational process, as, for instance, in an academic
institution. Typical players in such a scenario are the
educational service provider, the learners, the content
provider (e.g., a digital publisher,
o
the educational
institution itself), the course authors, the lecturers, the
tutors, and the admin staff members.
The GESTALT architecture thus consists of a set of
modular components envisaged as capable of supporting
the complete functional model. The basic components
are
:
the web client: in the GESTALT vision, user services
are delivered to the desktop using the WWW
technology;
the Learning Environment (LE): this is the core
component that, utilising an internal database,
provides on-line support for the typical actions
performed by the several players involved in the
educational process: curricula definition and course
construction, students enrolment, delivery of learning
resources to the desktop, tutor’s tracking of individual
student progress;
the Asset Management System: controls access to
valued resources, learning objects (i.e. course/module
data) which should only be accessed within a paid for
programme of study,
the Administration Management Information Services:
supports back-of-house management of the
educational/training institution.
One additional component of the functional model,
usually not present
in
distance learning systems, is the
Broker. This is an on-line entity that supplies information
about learning objects to users (learners, course authors,
etc.) wishing to discover them. The two related innovative
components provided in the GESTALT architecture are:
the Resource Discovery Service
(RDS):
it
allows Web
users
to
explore what courses and modules are
available from which institutions; the
RDS
implementation is based on the CORBA distributed
objects technology;
the User Profiles service: an LDAP-based directory
service used
to
store user preferences for LE
interaction (e.g. configuration options),
RDS
settings
(e.g. frequently searched hosts/services), and for smart
searches.
Within the project, big emphasis has been put on the
adoption
of
an open and standard data model underlying
the overall architecture, the
RDS
and user profile services
in particular. Emerging standards in the metadata field
have been explored and enhanced for this purpose. This is
described in the next section.
3.
The metadata model
One of the problems of modern education is the
overwhelming increase of information being constantly
produced and made available in new forms through the
WWW in various digital formats. The pace is such that
information is difficult to classify and some of it will
never be accessed by many potential learners.
A
vast
amount of data on the World Wide Web doesn’t even
hold classification information, which counts for difficult
and often imprecise searches. A number of initiatives, see
[I31
and [14] are trying to address the need for new and
standardized cataloging systems, which will enable to
share and access large amounts
of
digital resources across
institutions. All the initiatives use a set metadata for the
description of the resources. Some,
[16],
also propose
enhancements
to
the use of metadata.
Metadata are data about data. They provide a means of
describing a general type of data in a structured fashion
for use by many diverse applications. In the educational
field, like in many others, metadata facilitate
searchability, extensibility, re-usability, and scalability. In
an open scenario, essential is the possibility of querying
and exchanging metadata sets between heterogeneous
systems.
Work in GESTALT has lead to identifying three key
areas where the development of a standard metadata
model could allow achieving the above goals. The
complete GESTALT data model comprises the following
components:
-
Courseware Content (GEMSTONES);
-
Student Profiling/Tracking (PAPVEPAPI);
-
Curriculum Management (UOM).
The design of the three data models has drawn upon
the work of a number of standardization activities in the
field of metadata for education. Attention has focused on
the work done under the IEEE LTSC
[3],
the IMS project
[4], the ARIADNE project
[51
and the Dublin Core
[6].
Although not an educational initiative, the Dublin Core
aims at defining a set
of
cross-domain metadata to
enhance the searchability of electronic resources, and as
such it has been the foundation onto which other domain-
specific initiatives have been developed. The Dublin Core
has recently formed the DC Education work-ing group and
the first working draft is expected by end of February
2000.The final GESTALT data model is based on the
IEEE LOM v2.5 adopted by both IMS and ARIADNE.
XML’s DTD has been adopted for the binding of all
three data models and XML has then been used for the
creation of the metadata.
The first data model is GEMSTONES (Gestalt
Extensions to Metadata STandards for ON-line Education
Systems). It extends the LOM model, which uses just
87
eight categories to group the various metadata fields, by
adding three additional categories (shown
in
italics):
-
General
-
Lifecycle
-
Meta-metadata
-
Technical
-
Educational
-
Rights Management
-
Relation
-
Annotation
-
Assessment
-
Mappings
Figures la and Ib show the GEMSTONES DTD as
-
QoS
’
implemented
with
the Microstar’s Near
&
Far Designer.
GEMSTONE
GEMSTONE
Fig.
la: The GESTALT metadata model.
L
GEMSTONE
Fig.
1
b:
The
GESTALT
metadata model (cont.)
The Assessment category was introduced to describe
assessment-type learning resources;
it
is based on the
initial IMS model and
it
has been modified to bring
it
in
line
with
the LOM model. The
IMS
has recently formed a
Question and Test team (Q&T) with the scope of
developing a thorough Assessment data model
[2].
The Quality of Service (QoS) category (bottom of Fig.
Ib) was introduced to map a learning resource
requirements to the capabilities of the networking
technologies andor services necessary for its delivery.
Work has been undertaken at University
of
Naples to
define this category based on the IETF Integrated
Services/ Differentiated Services models, and on the basis
of the experiments
of
QoS delivery of audiolvideo
learning resources (described
in
section
4).
An important part of the GESTALT project has been
to demonstrate the effectiveness of “smart” types of
Internet searches when combining the student profile with
the metadata used for
the
description of the resources.
When using ordinary Internet search engines, a search
88
with a given keyword, say C++, will return thousands of
hits. Typically, only
a
small subset of the hits will be
relevant for the end student who, for instance, only speaks
English, is looking for an on-line C++ course and
has
certain preferences over the media format the course is
offered with, e.g. just text based material. By combining
the relevant keywords with the additional information
stored in the student profile, the search results will be
more accurate and also less numerous; this allows actual
manageability.
However, equivalent information within the learning
resources and the student profile will probably be
represented by different element names. The Mapping
category (bottom of Fig. lb) is then necessary to map
between different element names across the two data
models. In the GESTALT data model we only mapped
the student profile’s preferences (PAPI based) onto the
resources metadata elements’ name. The mapping
category can be similarly extended to allow searchability
over different data models such as the Dublin Core (DC),
provided that the DC-equivalent element is found in the
GEMSTONES structure.
The Student Profile data model draws upon the PAPI
work by Farance and Schoening
[7],
who aims at
facilitating the “interchange and interoperability of
student records, i.e. portable student records”. The
GESTALT data model extends this work by envisaging
two sub-models; the first, based on the PAPI work, allows
storing more static information about the student. It
includes learner information (Personal category), learner
accomplishments (Portfolio category), learner history
(Performance category) and learning styles (Preference
category). In the GESTALT architecture these
information are stored in an LDAP directory server and
regards a more dynamic type of information necessary for
tracking student progression in hisher learning
experience. Typical information includes the program of
study undertaken by the student, the organization
she
is
currently studying at, marks scored etc. The EPAPI data
model is use by the Learning Environment for tracking
student progress and enrolment information.
The third and last data model (Fig.
2)
is the Unit
Object Model (UOM), used between the Learning
Environment and the Management Information System
(MIS) for curriculum management. Typical information
in the UOM includes name of the unit, schedule for the
unit, whether it is compulsory
or
not, etc.
Figure
3
shows the relationships between all the above
data models and the components of the GESTALT
architecture.
EPAPI
Client
Content
Fig.
3:
The GESTALT architecture and data
models.
Similar work to GESTALT, with regard to intelligent
searches of educational resources combined with user
definable profiles, has been undertaken under the
Courseware Description (CDL) initiative[
111.
Some
3
ld-
0
7
startoate
He]
7
EnaDaw
+%Scheduleschema >Schedule
f+
7
Id
-
I1
L
Unit
-
f
+*
Occurrence
t
Fig.
2:
The Unit Object model metadata category
are accessed by the
RDS
for the “smart” searches. The other projects have also worked extensively to develop
second sub-model, called EPAPI (Extended PAPI) educational applications of metadata. The vast majority
of
89
data models and applications are mainly based on the DC
data model. Examples include the Gateway to
Educational Materials (GEM) [IO], the Education
Network Australia (EdNA)
[
181, the Virtual European
School (VES)
[
191,
and the European Schoolnet (EUN)
[20]. Numerous are finally the number of metadata-
standardization related initiatives and projects currently
running
[
151.
As for GEMSTONES, the main choice for
implementing the data models has been the XML DTD. A
major work is currently being done by the XML Schema
Working Group [I71
within
the W3C to develop
alternative and more powerful binding mechanisms,
which
will
allow for data typing and other important
features, see [12]. The EASEL project [21], on which the
authors are working at the moment, and which builds
upon the results of the GESTALT project, is considering
adopting an XML SCHEMA
in
place of the DTD.
4.
The GESTALT user
trials
The GESTALT user trials take place
in
the current
semester at University of Naples Federico I1 (UoN), and
they involve both members of staff and students,
representative of the different roles played by the users of
the GESTALT architecture:
lecturers (professors and assistant professors), acting
either
as
.tutors,
or
=course authors
(or
both);
0
a cohort of students;
0
a staff member, acting as administrator for student
The trials platform (Fig.
4)
consists of a number of
client computers and two server computers, installed
in
one of the laboratories of the Department of Computer
Science (DIS), running the various software components
providing the GESTALT architecture services: Resource
Discovery Service
(RDS),
Learning Environment (LE),
Management Information Services (MIS), Directory
Service (LDAP), etc. These servers are accessed by the
students from client computers
in
a different remote
laboratory, and by the tutors from their offices. The
connections are provided by the campus network. User
access through the Internet is supported too at the
addresses
http://servgest.grid.unina.it.
The trials are centred around the courses of
Fondamenti di Informatica I (Fundamentals of Computer
Science
I,
FII), and chiefly of Fondamenti di Informatica
I1 (FI2), belonging to the curriculum “Computer
Engineering” at the Faculty of Engineering. The latter
course runs over the first semester of the academic year,
from September to January.
applications and enrolment.
U
U
Fig.
4:
The
GESTALT
trials platform at
UoN
The first class of trial scenarios concerns the search
activities traditionally performed by students
or
course
authors, seeking a particular educational resource,
generally through a “standard” web search. In a normal
scenario, a user, be (s)he a student
or
a professional
navigating the web looking for an educational resource,
or
a course author looking for a resource for building his
own course’s set of resources, performs the search by
means of currently available Internet information retrieval
tools, getting a long list of often un-related results. With
the GESTALT brokerage services, based on the
RDS,
the
search is performed on a distributed database of metadata
descriptions of educational resources, provided by the
resource authors. Students and lecturers at UoN
experiment the effectiveness of the search looking for
learning material related to the topics of the F12 course;
the educational materials are mainly borrowed from the
previous EU RENAISSANCE project, with additional
multimedia material made available to the project for a
richer set of trials. The metadata accompanying the
resources have been developed with the technologies
(XML) and
in
the format (GEMSTONES) that were
defined
in
the previous phases of the project. A variation
of this search scenario is based on the user’s profile
stored on the LDAP server, used to refine the search.
The second category of trial scenarios concerns
activities in the course delivery phase. They are centred
around the use of the LE and
MIS.
A
tutor
uses
it
to
define the structure of the course F12 and link educational
resources, available
in
a variety of electronic formats, and
also to track students’ progress; a selected number of
students of the F12 course use the LE to access
courseware material interactively and interact with the
tutor asynchronously and remotely. A member
of
staff
cooperates with the tutor for trialling processing of
students applications and their enrolment. Fig.
5
(left)
shows a snapshot of the LE web page with a hypertextual
course
on
C++
programming delivered on-line.
90
The third category of user trials concerns the delivery
of multimedia courseware with QoS guarantee. These
trials aim at showing how metadata for resources can be
exploited, along with LDAP user profiles, to provide the
information necessary for multimedia applications to be
able to negotiate session parameters on networks with
QoS capabilities.
The multimedia resources used in these trials are pre-
recorded videoclips of lectures from a distance learning
course (provided by the italian NETTUNO distance
learning consortium
[SI)
covering some of the topics of
the
F12
course. The lectures (initially in VHS format)
have been digitalized and produced in standard digital
audiohideo formats
so
as to be delivered on-line through
the Web; a characterization of these videoclips has also
been produced, in terms
of
the network resources (e.g.
bandwidth, peak rate) required for quality on-line
delivery. The DiVA video-on-demand prototype
developed at DIS is used for these trials on a controlled
QoS network testbed in the DIS laboratory. Fig.
5
(bottom) is a snapshot of the client desktop with the LE
window and DiVA client window playing a recorded
lecture.
5.
Results
The user trials involve one professor and one assistant
professor (co-authors of this paper), also acting as tutors
for the trials, one admin staff member, and a subset of
about twenty students of the two selected institutional
academic courses. The preliminary results of the
educational activities centred around the LE, and the
digital delivery of courseware.
As for the RDS retrieval support system, the user tests
performed covered mainly functional aspects. About
80
metadata descriptions were built for learning resources
and inserted in the search database of the
RDS.
The
results confirm the high potential usefulness of the
service, especially in its interoperability with the LE-
based delivery service. However, the tests do not allow
to
fully assess its effectiveness. Further tests would be
required, with a larger metadata sets and in a real-world
scenario, to definitely assess the benefits of the “smart”
searches with respect to traditional web searches.
To the aim of the trials, the tutors compared the use of
the
FIl
and
FI2
existing web sites with the LE-based
approach.
If
on the one hand, appropriate training was
required that allowed tutors to use the LE in all its
functionalities, on the other hand the tutors felt somewhat
lightened from their traditional cumbersome Web
management duties. Although competent in the Web
technologies, a tutor can often find these duties
overwhelming and difficult to cope with. The
functionalities offered by the LE proved to be
advantageous over the traditional “web-siting” of course
notes and information. Very useful was the student
tracking capability.
The trials with students have been followed by a
system usability assessment phase, based on the
methodology and tools developed and provided by project
partners at the Trinity College Dublin
[9].
Students and
tutors had access to a web site at TCD to fill user
Fig.
5:
snapshot
of
the student’s view
o
f
the
LE
with a
hypertextual (left) and a multimedia (right) learning resource
experience made allow to draw comments on three key
aspects of the overall business process envisaged in
GESTALT: the learning resource search service, the
questionnaires and then analyse results, respectively.
Questions aim at assessing usability aspects, such as
friendliness and layout consistency of human interfaces of
electronic instructional materials.
91
More specifically,
5
criteria are considered by the
assessment methodology: A) naturalness, B) user support,
C) consistency, D) non-redundancy, E) flexibility. A
sample question for criterion D is:
“Is
there an adequate
amount of instructional. material on each screen?”.
Preliminary results based on the analysis of the
questionnaires on the courseware material delivered
through the
LE
are summarised in Table
I.
Table
I:
Results
of
the usability assessment
Agree
68,7
1
01
E
0
16,7 83,3
0
Finally, as for the QoS guaranteed delivery of digital
resources, the trials on the UoN laboratory testbed
demonstrated the effectiveness of the approach of
integrating metadata and user profiles technologies with
multimedia applications and network protocols. The cost
for this is that of characterising the multimedia resource
in terms of (a limited set
of)
parameters expressing the
requirements on the network (and possibly on the client
computer), to be negotiated with QoS-aware network
equipment. The more precise the media characterisation,
the more effective the usage of network resources can be.
This characterisation
is
probably to be considered an
authoring issue, but
it
is unlikely to be left up
to
the
content authors, since it requires advanced technical
skills.
One further result of the QoS experiments
has
been the
refinement of the QoS schema within the GEMSTONE
metadata model. The detailed description of this work is
beyond the scope of this paper; it is reported in reference
PI.
6.
Conclusions
In this paper we have given an overview of the
architecture and data model of the GESTALT system for
supporting on-line retrieval, management and delivery of
learning resources, and we have described the experience
done at
UoN
setting up and using the system with a
subset of students of two institutional courses of
Computer Engineering
.
One of the main achievements of the project has been
to show how a number of different technologies,
including web, brokerage, metadata, digital authoring and
network (QoS) protocols, can be integrated
to
provide
flexible support for educational activities in an open
scenario. The project succeeded in this respect, and it was
judged just one point off the maximum possible score by
the auditors.
The trials and the results presented in this paper have
confirmed this approach and encourage to further explore
these issues.
Acknowledgements
The work presented in this paper is part of the EC-
funded GESTALT project, which has seen the
participation of research groups in several academic and
industrial partners, namely: Fretwell-Downing Education
(UK, group leader K. Riley, also manager of the project),
British Telecom (UK,
P.
Foster), Kyros (GR, D.
Maroulis and R. Smith), Trinity College Dublin
(IRL,
V.
Wade), University of the Agean (GR,
J.
Dartzentas),
University of Naples
(I,
S.
Russo and G. Ventre) and
Waterford Institute of Technology
(IRL,
W. Donnelly).
References
[I]
The GESTALT
Project,htto://www.fdgrouo.co.uk/gestalt/
[2] P. Foster
et
al,
The GESTALT project, Courseware
Metadata Design (GEMSTONES), Project Deliverable D401,
h
ttp://www.fdgroup.co.uk/gestalt/D0401
.pdf
[3] IEEE Learning Technologies Standards Commi ttee (LTSC),
h
ttp://grouper.ieee.org/groups/ltsc/
[4]
The
Instructional Management Systems
(IMS)
Project,
http://www.imsproject.org/
[
51
ARIADNE Educational Metadata Recommendation
Summary, Working document, Version 2.0, April 1998
[6]
Dublin Core Metadata for Resource Discovery. IETF RFC
241 3
[7] F. Farance,
J.
Schoening, PAP1
-
Personal And Private
Information.
http://edutool.com/papi/
[8]
The NEITUNO Distance University Network,
http://l95.62.32.250/main.html
[9] V. Wade,
M.
Lyng, Experience Improving
WWW
based
Courseware through Usability studies, AACE World
Conference on Educational Multimedia and Hypermedia
&
World Conference on Educational Telecommunications 99,
Boston, 1999.
[IO]
Stuart
A.
Sutton, Gateway to Educational Materials
(GEM): metadata for networked information discovery and
retrieval, Computer Networks and ISDN Systems, Volume 30,
hues 1-7, April 1998, pp. 691 -693
[l
I]
L. Jay Wantz, M. Miller, Educational metadata
in
use.
Computer Networks and ISDN Systems, Volume
30,
Issues 1-7,
[I21
D.
Ioannides, XML schema languages: beyond DTD.
Library
Hi
Tech, Volume
18,
Number
1,
pp 9-14.
April 1998, pp. 701-703
92
[I31 D. Domer, Cataloging in the 21st century
-
part2:
digitization and information standards. Library Collections,
Acquisitions,
&
Technical Services, Volume 24, lsue 1, Spring
[
141 J. Darzentas, Sharing metadata: enabling online
information provision. OCLC Systems
&
Services, Volume 15,
number 4,1999, pp. 172-178.
[I51 J. Milstead,
S.
Feldman, Metadata: Cataloging by Any
Other Name. ONLINE, January 1999
[I61 M. Marchiori, The limits of
Web
metadata, and beyond.
Computer Networks and ISDN Systems, Volume 30, Issues 1-7,
[
171 XML Schema Working Group. htt~://www.w3.org
[
181 Education Network Australia. htto://www.edna.edu.au
[
191 Virtual European School.
htto://www.ves.eu.orn/home.htm
[20] European Schoolnet.
htto://www.en.eun.ordfront/actual/index.
html
I211 Educator Access to Services in the Electronic Landscape
(EASEL
-
IST Project
10051).
htta://www.fdnrouo.co.uk/easel/
2000, pp. 73-87
April 1998, pp. 1-9
93