SPECIFICATION OF OBSERVATION NEEDS IN AN
INSTRUCTIONAL DESIGN CONTEXT
A Model-Driven Engineering Approach
Pierre Laforcade, Boubekeur Zendagui and Vincent Barré
Maine University - LIUM
IUT de Laval, 52 rue des Docteurs Calmette et Guérin, 53020 Laval Cedex 9, France
Keywords: Instructional Design, Learning Scenarios, Observation, Model Driven Engineering, Meta-Modelling.
Abstract: Our works take place in the research field of distant learning situation observations. We want to help
instructional designers to improve the learning scenarios they design within a re-engineering context. We
think that the observation of the learners' behavior can be improved by taking into account the observation
needs from the designers. We originally think these observation needs can be related to and guided by the
information specified into the learning scenarios. This article presents a Model-Driven Engineering
approach for the specification of these observation needs. A specific metamodel has been elaborated to
support our conceptual proposition and process. A dedicated example illustrates the use of this metamodel
to specify observation needs according to a given learning scenario and Educational Modeling Language.
First elements of tooling are also presented.
1 INTRODUCTION
In the research field of distant learning situation
observation, most of the research deals with the
analysis of the collected data during the learning
session. Our works take place within a re-
engineering context of these learning situations.
We want to help instructional designers to
improve the learning scenarios they design. To
achieve that, we assume and believe that the
observation of the learners' behavior could be
improved by taking into account the observation
needs from the designers a priori of the learning
session.
We also originally think that these observation
needs can be related to and guided by information
from the learning scenarios specifying the various
facets of the learning flow. We aim at supporting
and guiding designers when defining these
observation needs. To this end we have already
proposed a conceptual model for the observation
needs, whose originality relies on the use of 'signs'
and 'behavior categories' techniques
(Zendagui et al.,
08)
as possible features for detailing observation
needs on the base of learning scenario information.
This paper focuses on the Model-Driven
Engineering context and approach we follow to
propose both a theoretical formalization of the links
between observation needs and learning scenarios,
and a practical tool for helping designers to define
these needs. We follow such an approach because
our previous work about Domain-Specific Modeling
(DSM) & Educational Modeling Languages (EML)
(Laforcade et al., 08) focuses on the use of meta-
modeling techniques and Eclipse tools to support our
proposition as a first formalization step.
The next section presents our research context
about instructional design and observation needs. In
the following section we briefly present our
conceptual proposition. The MDE approach for the
support of our conceptual proposition is then
detailed and discussed in section 4. Then, section 5
is dedicated to our first results illustrating the use of
specific MDE tools for supporting our metamodel
and providing a tree-based graphical editor. Finally
we conclude by discussing our research in progress
and the benefits designers can expect to obtain.
111
Laforcade P., Zendagui B. and Barré V. (2009).
SPECIFICATION OF OBSERVATION NEEDS IN AN INSTRUCTIONAL DESIGN CONTEXT - A Model-Driven Engineering Approach.
In Proceedings of the First International Conference on Computer Supported Education, pages 111-118
DOI: 10.5220/0001978901110118
Copyright
c
SciTePress
2 INSTRUCTIONAL DESIGN
CONTEXT
2.1 Re-engineering of Learning
Scenarios
Within the instructional design context, the re-
engineering of learning scenarios forms a cycle. In
the first step, designers use an Educational Modeling
Language (EML)
(Koper et al., 05)(Kinshuk et al., 06)
to define a learning scenario. This predictive model
allows designers to explicit their pedagogical
objectives regarding learning situations in terms of
activities/tasks, roles, resources/services,
objectives/prerequisites, etc.
In the next step, the concrete learning situation is
in progress, and leads students/tutors (and other
associated actors) to use the designer's pedagogical
scenario. In this step, data regarding effective use of
the learning situation is collected either by the
Learning Management System (LMS) or by other
means. Many experiments have shown that the
effective running of a pedagogical situation does not
necessarily follow the predictive scenario, leading to
the need to observe the real behavior of the actors.
To be used in a learning scenario re-engineering
process, this data must be analyzed and abstracted
from the system format that produced it ;this
facilitates its interpretation by the analysts and the
instructional designers. So, the last step of this cycle
consists of analyzing data collected during or after
the effective running of the learning situation. This
data is then converted (filtered, structured,
combined, etc.) leading to enriched data, having a
pedagogical meaning for the designer.
The results of this step then allow designers to
compare the predictive scenario with the observed
situation. Thus, they can be brought to modify their
predictive scenario as a first step of a new iteration.
2.2 REDiM Project
Our research works take place within the REDiM
project
(Choquet, 07) whose main objective is to
provide teachers with dedicated techniques and tools
supportingt the re-engineering of their learning
scenarios. Within this project, the UTL language has
been proposed for the XML specification of
observation datum / observations means and
observation needs
(Choquet et al., 06). However there
is still a lack of learning-scenario-centered tools and
practices to help designers in specifying their
observation needs.
Our current research focus on this lack. To our
minds, the preparation of the learning situation
observation is an activity that relies on the one
dealing with the design of learning scenarios
(Zendagui et al., 08). We aim at helping designers in
defining what is important to observe during the
elaboration phase of the learning scenario. This
process must deal with many potential difficulties
for the expressiveness, relevance and usefulness of
observation needs:
z they are linked to the learning scenario
expressiveness and, by extension, to the
expressiveness of the underlying EML;
z they depend on the designers' ability to specify what
they want to observe and which information they
need;
z they have to be specified with such details and
formalized in such a machine-readable format that it
will be able to automatically handle them to, for
example, guide and help the track analysis.
We think that the MDE approach can formalize and
help us in tackling these issues.
To assist designers in their observation needs
specification task, we have studied the observation
activity and its preparation within both classic face-
to-face and distance learning situations
(De Ketele,
87)(Wragg, 99)(Dessus, 07)
. The next section presents
our conceptual proposition.
3 THE CONCEPTUAL
PROPOSITION
3.1 The Process
To assist designers in their observation needs
specification task, we propose a two-step approach
(see figure 1).
3.1.1 Observable Identification
Within this process, the effective EML used by
designers is not a priori known. It can be improved
or enriched to better express designers needs
(improvement of the EML expressiveness) during its
use. We propose to identify the elements (concepts,
relations, attributes) of the EML that can potentially
be observed, tagging them as “observable”. We
define an observable as any EML element whose
change could be meaningful to observe; i.e. any
element whose instantiation (for a concept), value
(for an attribute), or acquaintances (for a r elation)
CSEDU 2009 - International Conference on Computer Supported Education
112
Figure 1: Teacher/designer-centred process for specifying
observation needs.
can be useful to observe. For example, one might
want to observe how many times an activity will be
carried out (the “Activity” concept can thus be a
possible observable), or to observe the different
values of a “duration” property on a specific
activity, or to concretely observe the order of an
activity sequence (thus, identifying as observable the
precedence” relation defined on two activities).
This categorization of potential observables has
to involve the instructional designers, or other
experts of the pedagogical domain, in order to be
sufficiently relevant. This identification of
observables is, a minima, human-directed but it will
be possible to eventually guide this activity thanks to
a semi-automatic process based on a suggestion of
observables according to a syntax-and-semantics
analysis of the EML
(Barré et al., 05).
3.1.2 Observation Needs Specification
The second step is the concrete specification activity
of observation needs. Designers must precisely
specify the observation needs they consider as
relevant. Those observation needs will be intimately
specified in relationship with a specific learning
scenario. This activity can be done after the scenario
specification or concurrently.
For each definition of an observation need,
designers will follow three sub-activities: definition
of the context, selection of observables, and
definition of signs or categories of behavior.
The context definition aims at “contextualizing”
the observation need. In addition to the observation
objectives, designers have to define a subset of
learning scenario elements concerned by the
observation objectives. Thanks to this delimitation,
only the observables concerned by elements within
the context are selected and proposed to designers
for the next sub-activity. The designers then have the
responsibility to decide to use, or not, the proposed
observables to clarify which data must be collected
concerning the future learning situation. After this
selection, designers can then define signs or
categories of behavior (detailed in the 3.2
subsection) on these observables
(Wragg E.C 99).
If the selected observables are not useful,
designers can go back to the observable selection
phase, as well as the context definition phase, in
order to add new observables or scenario elements.
They can also modify the pedagogical
expressiveness of the EML, as well as modifying the
observable identification at the EML level.
3.2 The BOSIC Conceptual Model
We define an observation need as composed of four
parts (see figure 2): observation objectives,
observation context, elements to observe or
observables, and signs and behavior categories
defined on one or more observables.
The observation objectives are useful to define
the “why” of observation needs. This information
allows designers to explicit what they want to do
when they know the results of their observation
needs from the concrete observation of learning
situation runtime. This information can also be
useful to facilitate the reusing of observation needs
for other learning situations or other learning
scenarios sharing the same objectives.
The observation context allows the definition of
conditions under which an observation need is
defined and used. Contexts allow the delimitation
the learning activity to observe. It consists of
selecting one or more pedagogical scenario
elements. Contexts are important and must be well
defined since they allow the identification of the
potential observables.
Figure 2: The BOSIC conceptual model.
The observables are pedagogical scenario
elements for which designers want to get
information after the learning situation execution.
Concretely, these observables are defined at a
scenario level butconform to those defined at the
SPECIFICATION OF OBSERVATION NEEDS IN AN INSTRUCTIONAL DESIGN CONTEXT - A Model-Driven
Engineering Approach
113
EML level: they are contextualized and correspond
to a kind of 'instantiation' of the EML-level
observables with respect to their context. Their
specification is done by selecting observables among
those than can be automatically proposed according
to the context delimitation and the observables
identified at the EML level. For example, “duration
of activity X” is an automatically proposed
observable since “activity X” is part of the
observation context and since the “duration”
attribute of the “activity” concept is identified as an
observable at the EML level.
The additional data is defined on one or more
selected observables at a scenario level to
characterize a first representation of the observation
means, in a domain-oriented language, identified by
the experts. The signs correspond to particular
events for which designers want to know apparition
frequency, occurrence number or, a minima, to
know if they occur or not, during a future
observation. One or more signs can be defined on
one or more scenario-level observables. For
example, the “duration of at least 30 minutes” sign
can be defined on the observable (“duration of
activity X”). The behavior categories correspond to
some grouping of events that will be analyzed as a
block. Like the signs, one or more categories can be
defined in relation to one or more observables at the
scenario level. For example, the “handling events”
category can be defined to indicate that we want to
know all the undergone manipulations observed on
the “document Y”; on the condition that the
“document Y” is a selected observable, and by
definition is also part of the delimited context, and
that the underlying metaconcept (“Deliverable” for
this case) was identified as an observable at the
EML level.
The next section is about the formalization of our
conceptual propositions following the MDE theories
and practices.
4 THE MDE APPROACH
The formalization of the BOSIC conceptual model is
necessary for two goals: firstly, to concretely
support our proposition and propose a machine-
readable notation for the observation needs (that will
be useful, by extension, to manipulate/transform the
specified needs to facilitate their reuse, etc.), and
secondly, to help the building of software tools to
support the observation needs specification activity
by designers (user-friendly editors with helpful and
guiding facilities, etc.).
4.1 MDE Projection
The Model Driven Engineering (MDE) is a software
development methodology which focuses on
creating models that describe the elements of a
system
(Schmidt, 06). A modeling paradigm for MDE
is considered effective if its models make sense from
the point of view of the user and can serve as a basis
for implementing systems (productive models). The
MDE principles (abstraction, modeling, meta-
modeling, separation of concerns, etc.) have been
applied within various educational disciplinary
fields: adaptable learning materials generation,
Computer Supported Cooperative Work, etc.; we
have studied the application of its theories and
practices for learning-scenario-centered instructional
design processes in
(Laforcade et al., 07)(Laforcade et
al., 08).
From a MDE point-of-view, and if we do not take
into account notation or concrete syntax aspects, the
learning scenario is a model conformed to the
metamodel specifying the terminology, or abstract
syntax, of the EML used to define the scenario (see
figure 3). Similarly, a set of observation needs,
defined once for a given learning scenario,
corresponds to a model. Because we want these
observation needs defined in relation to information
from the learning scenario, the models have to be
linked together too. The model specifying some
observation needs also has to be conformed to a
specific metamodel for the definition of observation
needs: the BOSIC metamodel (detailed in the next
sub-section). This metamodel has to refer to the
information from the EML metamodel.
Figure 3 is a four-layer OMG
(OMG, 06)
representation of these MDE artefacts.
Figure 3: OMG layers view of the observation needs.
CSEDU 2009 - International Conference on Computer Supported Education
114
4.2 MDE Techniques Used
According to the conceptual process we outlined in
the previous section, and to the MDE practices, we
need concrete techniques to support both our
conceptual process and BOSIC model.
The BOSIC metamodel as well as the EMLs used
to define learning scenarios have to be specified
using the meta-modeling technique: construction of
a collection of "concepts" (things, terms, etc.) within
a certain domain
(Wikipedia, 08). A metamodel is a
precise definition of the constructs and rules needed
for creating semantic models. We illustrate an
example of a metamodel for an EML in section 5,
whereas the BOSIC metamodel is detailed in section
4.3.
We also need a technique to add information
about the elements that designers want to tag as
'observable' to the EML metamodel . Because these
potential elements can be concepts, attributes, as
well as relations between concepts and because
every metamodel conforms to the unique meta-meta-
model MOF
(OMG, 06), we need to use a MOF
concept to be able to attach the 'observable'
information to the class, attributes, association, etc.
This meta-construction is called annotation (for the
MOF 1.4), or comment (for the MOF 2.0). Section 5
shows how we use the equivalent EAnnotation
mechanism from the Eclipse Modeling Framework
(EMF) tooling.
Another issue from our conceptual process and
model is how can the 'context' and 'observable' parts
of an 'observation needs model' can refer to elements
from a specific 'learning scenario'. Following the
fact that the specification of an EML metamodel can
use various MOF building blocks (class, attributes,
relations...), it is not possible to specify a meta-
relation in the BOSIC metamodel to 'anything-
specified' in the EML metamodel. Also, because
EMLs can differ, it is more relevant to concretely
separate them. From our MDE expertise we choose
to add in the BOSIC metamodel a specific concept
which plays the role of a 'proxy' for the learning
scenarios elements (see the next subsection).
4.3 The EMF Tooling
To concretely formalize and support the
development of our proposition and dedicated
editors, we chose to use a unified set of modeling
frameworks, tooling, and standard implementations
from the Eclipse Modeling Projects
(Eclipse EMP,
08): EMF, GMF and ATL. In this article we only
focus on the Eclipse Modeling Framework (EMF)
because it provides the very first layer of support we
need.
Indeed, the EMF is a modeling framework and
code generation facility for building tools and other
applications based on a structured metamodel
(Steinberg et al., 2008). From a metamodel
specification, EMF provides tools and runtime
support to produce a set of Java classes for the
metamodel, along with a set of adapter classes that
enable viewing and command-based editing of the
model, and a basic editor.
We illustrate the use of this very basic editor in
section 5. Also, we have planned to use the
Graphical Modeling Framework (GMF) in a second
time to add a graphical layer on top of EMF, and,
incidentally, to develop a graphical editor dedicated
to the specification of observation needs.
Finally, the ATLAS Transformation Language
(ATL) is the model-to-model transformation
framework we will use to transform observation
needs conformed to our BOSIC proposition into
other machine-readable formats for the specification
for observation needs (like the XML-based one
proposed by UTL).
4.4 The BOSIC Metamodel
This sub-section details the BOSIC metamodel we
have specified using the EMF tooling (metamodels
are called ECORE models where ECORE is the
MOF-like meta-meta-model in EMF). Figure 4
illustrates this metamodel in the class-diagram-
oriented view proposed by the Ecore graphical
internal editor of EMF.
The ObservationNeeds concept plays the role of
the root for the specification of several observation
needs in a same model, according to a same learning
scenario too. In addition, the ObservationNeed
concept is the root node for all the information in
regard to one observation need. It is composed of
several Objectives and of three other concepts
representing the three layers of an observation need:
ContextLayer, ObservablesLayer and DataLayer.
The ContextLayer concept is the node element
under which are specified the PSElements (scenario
elements). As previously explained, this EClass is a
kind of 'proxy' that refers to an element from the
learning scenario: it can be an instance of an EClass
from the EML metamodel, a property (the
EAttribute and value pair) for a specific instance of
an EClass, as well as a link (instance of a ERelation)
between two instances of EClasses. For our very
first prototype, these PSElements will have to be
specified as new inputs even if they already exist in
SPECIFICATION OF OBSERVATION NEEDS IN AN INSTRUCTIONAL DESIGN CONTEXT - A Model-Driven
Engineering Approach
115
Figure 4: the BOSIC metamodel.
the learning scenario but future versions will provide
designers with guiding facilities.
Similarly, the ObservablesLayer gathers the
SelectedObservable which represents, according to
the specification of the designers, a subset of the
DeclaredObservables, the observables at a scenario
level that can be deduced from the PSElements
previously specified, and from the EML metamodel
elements tagged as observable. We also plan to
develop a specific algorithm and code routines
dealing with the DeclaredObservables to
automatically instantiate them with the deducible
information.
Finally, the ObservationTechniques concept
allows the definition of signs and categories of
behavior (inheritance relation). It represents the third
layer of an observation need. This information
isdefined using one or more SelectedObservable via
the informationOnObservable relation.
5 ILLUSTRATION
To illustrate our propositions as well as the first
prototype we developed, we now present and discuss
a concrete example of specification of observation
needs according to a learning scenario and its
underlying EML.
5.1 The EML Metamodel
Among the various case studies we have
experimented on with EMF and GMF, we outline
the following one for this article. Some practitioners
have expressed these pedagogical expressiveness
and notation needs: a UML UseCase-like diagram
that shows the performing relations between roles
and learning activities at a high-level of abstraction,
and precedence/following relations between learning
activities. We have therefore provided them with a
specific graphical EML (or VIDL for Visual
Instructional Design Language)
(Botturi et al., 07) and
a dedicated visual editor using EMF/GMF. A
metamodel for the « Learning Design Use Case »
(LDUC) view has been defined. It is illustrated in
figure 5.
According to our conceptual process, designers
have identified these potential observables from the
terminology crystallized by this metamodel: some
observables are EMF EClass (eg. HighLevelActivity,
Actor), some are EMF EAttributes (eg. Duration,
score), others are EMF EReferences (eg.
nextActivities). All these elements have been tagged
as observables using the EMF EAnnotation
Figure 5: example of EML metamodel - the LDUC
metamodel.
mechanism (only the EAnnotations on EClass are
shown in figure 5).
5.2 The Learning Scenario
From the previous EML, designers have proposed
the following scenario (extract on figure 6) using the
graphic editor we developed thanks to the
EMF/GMF frameworks. Briefly, this scenario
focuses on the specific phase (“OS introduction”) of
a learning scenario they want to play. The first
learning activity “updating” is composed of a
sequence (specified thanks to the “next” and
“include” relations) of sub-activities “QCM”,
Answers consultation” and “exchange” (using a
forum).
One must know that the graphic representation of
the learning scenario does not reflect all the
information specified with the LDUC editor: some
have no graphic representations and can only be
seen in the properties view of the editor (eg. the
duration property for any activity).
CSEDU 2009 - International Conference on Computer Supported Education
116
Figure 6: An example of a learning scenario.
5.3 The Observation Needs Model
Knowing both this learning scenario and the
specificities of the distant platform (or LMS) that
will be used to play this scenario, the instructional
design team (where the instructional designers are
the experts for formalizing of pedagogical
intentions) want to know if the learners really take
the time to consult the answers and resources
available after the QCM activity, before using the
forum to get explanations from their peers. For
them, this question is meaningful because the LMS
leaves learners independent for the order in which
they perform learning activities and because the
LMS does not limit the time activity to strictly
follow what the designers specify using the duration
property.
This observation need aims at gathering
information about the use of the “Answers
consultation” activity. The most meaningful
information for this activity are that about learners
who fail the “QCM” activity with a low score (less
than 50%). It is the concrete objective of the
teachers.
Concerning the context of this observation need,
designers must select these elements from the
learning scenario: the activityanswers
consultation”, the “QCM” activity and the link
next” between the two activities. This context
filters the potential observables (Declared-
Observables) that can be proposed to the designers:
next”, “answers_consultation”, “answers_consul-
tation.duration”, “answers consultation.score”,
“QCM”, “QCM.duration” and “QCM.Score”.
From this list of potential observables, designers
only selected the following observables:
“answers_consultation.duration”, “QCM.Score”
and the next” link.
Designers then specified three observation
techniques: two signs and one category of
behavior. The first sign aims to give information
about the number of learners who consult the
resources/answers directly at the end of the QCM
activity. The second sign focuses on the number of
learners that score less than 50% for the QCM
activity and that spend less than one minute on the
answers consultation” activity. A large number
here will indicate that learners did not make an
adequate effort to understand their mistakes. Finally,
the category of behavior designers have defined
aims at collecting all the durations (for each learner)
for the “answers consultation” activity. This
expected set of results will give designers with the
time spent by each learner in the “Answers
consultation” activity.
5.4 The First Prototype
The first prototype has been developed thanks to the
EMF tooling. This framework generated a first
version of the editor directly from our BOSIC
metamodel. This editor provides designers with a
tree-view of the models where each node is an
instance of an EClass from the metamodel and child
nodes are the instances of EClass linked by a
relation of containment between the two EClasses in
metamodels (see figure 7). Properties and links (kind
of “instances” at a model-level of the EAttributes
and ERelation defined at the metamodel level)
appear in the property view according to the element
selected in the tree-view.
In addition to the Java-code generated by EMF
we have added some specific modifications to adapt
the editor: personalized labels in the tree-view,
interrogating routines to gather meta-information on
any elements from the learning scenario, and
generation routines to automatically instantiate
DeclaredObservable according to the potential
observables information that can be deduced. The
screen-capture depicted in figure 7 shows how we
used this editor to formally specify the observation
needs we used as illustration.
6 CONCLUSIONS
This article has presented and discussed a specific
Model-Driven Engineering approach for the support
SPECIFICATION OF OBSERVATION NEEDS IN AN INSTRUCTIONAL DESIGN CONTEXT - A Model-Driven
Engineering Approach
117
of the definition of observation needs as models in
relation to the 'learning scenario' model. We have
proposed a support that is both theoretical, by
providing a conceptual model and a specific process,
and practical, by specifying a dedicated metamodel
and by generating the first prototype of a dedicated
editor, according to the use of the Eclipse Modeling
Framework.
For now, we are working on several
improvements. Some are conceptual like the use of a
neutral referential for any constructivist-oriented
EML to ease the definition of the context for an
observation need. Other improvements are related to
our MDE approach and tooling: we want to improve
the editor prototype by dealing with concrete
syntaxe (notation) aspects for a graphic definition of
observation needs. To this aim, we have already
used Eclipse's GMF (Graphical Modeling
Framework) to provide practitioners with Visual
Instructional Design Languages (VIDL) and
dedicated editors. We plan also to use GMF to add a
graphic layer on top of the EMF-generated tree-view
editor for the specification of observation needs. We
think that this graphical layer will give us access to
facilities and services for more user-friendly editors.
REFERENCES
Barré V., Choquet C., 2005. Language Independent Rules
for Suggesting and Formalizing Observed Uses in a
Pedagogical Reengineering Context. In: ICALT'05.
July 5-8. Kaohsiung (Taiwan). pp. 550-554.
Botturi, L., Stubbs, T. (eds.), 2007. Handbook of Visual
Languages in Instructional Design: Theories and
Practices. Hershey, PA: Idea Group.
Chikofsky, E. J., Cross, II J. H., 1990. Reverse
engineering and design recovery: A taxonomy. IEEE
Software, 7(1).
Choquet, C., Iksal S., 2006. Usage Tracking Language: a
meta language for modelling tracks in TEL systems. In
International Conference on Software and Data
Technologies (ICSOFT). September 11-14
th
. Setubal
(Portugal). Filipe J., Shishkov B., Helfert M. (ed.),
ISBN 978-972-8865-69-6.
Choquet, C. 2007. Engineering and re-engineering of TEL
systems, the REDiM approach. Habilitation à Diriger
des Recherches. Le Maine University. France. In
french.
De Ketele, J.M., 1987. Méthodologie de l’observation.
Bruxelles, De Boeck.
Dessus, P., 2007. Systèmes d’observation de classes et
prise en compte de la complexité des événements
scolaires. In Carrefours de l'Éducation. Vol(23). pp.
103-118.
Eclipse EMP, 2008. Eclipse Modeling Projects.
http://www.eclipse.org/modeling/. Retrieved from
Decembre 2008.
Favre, J.M., 2004. Towards a Basic Theory to Model
Driven Engineering. In WISME'04, the 3rd Worksho p
in Software Model Engineering. Lisbon, Portugal .
Kent S., 2002. Model Driven Engineering. In IFM 2002.
LNCS 2335. pp. 286 – 298. Springer – Verlag.
Kinshuk, Sampson D.G., Patel A., Oppermann R., 2006.
Special issue: Current Research in Learning Design. In
Journal of Educational Technology & Society. (9)- 1.
Koper, R., & Tattersall, C., 2005. Learning Design – a
handbook on Modelling and Delivering Networked
Education and Training. Springer-Verlag. Berlin
Heidelberg.
Laforcade, P., 2007. Visualization of Learning Scenarios
with UML4LD. In: Journal of Learning Design. Vol.
2 (2). pp. 31-42.
Laforcade P., Zendagui B., Barré V., 2008. Supporting the
Specification of Educational Modeling Languages and
Learning Scenarios with a Domain-Specific-Modeling
Approach. In: ICALT'08. July 01-05. Santander
(Spain). pp. 819-821. IEEE Computer Society.
Metamodel, 2008. http://www.metamodel.com/. Retrieved
from December 2008.
OMG, 2006. Meta Object Facility specification (MOF
2.0). Formal/2006-01-01. http://www.omg.org/mof/
Schmidt, D.C., 2006. Model-Driven Engineering. IEEE
Computer. 39 (2).
Steinberg, D., Budinsky, F., Paternostro, M., 2008. EMF:
Eclipse Modeling Framework, Second Edition.
Publisher: Addison Wesley Professional.
Wikipedia, 2008. Meta-modeling.
http://en.wikipedia.org/wiki/Metamodeling. Retrieved
from December 2008.
Wragg E.C., 1999. Introduction to classroom observation.
Second edition. Routeledge, May 14.
Zendagui B., Barré V., Laforcade P., 2008. Support to the
specification of observation needs. In: ICALT'08. July
01-05. Santander (Spain). pp. 793-797. IEEE
Computer Society. ISBN 978-0-7695-3167-0.
CSEDU 2009 - International Conference on Computer Supported Education
118