Turning Recurrent Uses of E-learning Tools into Reusable Pedagogical

Activities

A Meta-modeling Approach Applied to a Moodle Case-study

Esteban Loiseau, Pierre Laforcade and Nour El Mawas

LIUM (Computer Science laboratory), University of Le Mans, Le Mans, France

Keywords:

Instructional Design, Visual Instructional Design Language, Learning Management System, Modeling and

Meta-modeling, Weaving Models.

Abstract:

The pedagogical expressiveness of designed courses using Learning Management Systems (LMS) is highly

dependent on the teachers’ expertise about how to use the LMS at their disposal. The GraphiT project aims to

help teachers in focusing on the speciﬁcation of pedagogically sound learning scenarios that can be technically

executable for automatically setting-up the LMS. We propose to provide teachers with LMS-speciﬁc instruc-

tional design languages and editors. The LMS semantics has to be raised in order to enrich the pedagogical

expressiveness of the produced models. This paper deals with the proposition of a speciﬁc LMS-centered

approach for abstracting the LMS low-level parameterizations and turning them into higher-level pedagogical

building blocks. We choose to present and illustrate our propositions about the Moodle LMS. We focus on

the ﬁrst abstraction level. It consists in identifying some pedagogical activities according to recurrent uses

designers make by handling the Moodle activities.

1 INTRODUCTION

Nowadays, Learning Management Systems (LMS)

are widely spread in academic institutions. These

LMSs are not restricted to distant courses but are also

useful during or in complement to face-to-face learn-

ing sessions (Garrisson & Kanuka, 04). Nevertheless,

the results of a study we conducted with 203 teachers,

who put forward their heavy form-oriented human-

interfaces and their tools/services-oriented course de-

sign, which lead to reduce their uses. In order to set

up complex learning activities, teachers must develop

high-level skills on how to use the existing LMS. Such

skills can be acquired through speciﬁc teacher edu-

cation programs, often focusing on the features and

technical aspects of the platform, but few courses deal

with how to design pedagogically sound learning sit-

uations for this very speciﬁc LMS. Because of the

multiple educational theories and approaches (Orm-

rod, 11), as well as the lack of tools and processes

dedicated to existing LMSs, teachers develop ad hoc

and individual learning design techniques.

In such context, it is relevant to help teachers in fo-

cusing on pedagogical aspects and their instructional

design set-up for the speciﬁc LMS they have at their

disposal. Whereas improving their knowledge and

abilities to use the platforms’ features, a focus on

the instructional design possibilities and how they can

rely on the platform features should encourage indi-

vidual and collective understanding about the peda-

gogical uses of the targeted LMS.

For this purpose, we propose an LMS-centered

design approach in opposition to the usual platform-

independent approaches (Alario-Hoyos et al.,

12)(Katsamani et al., 12). The GraphiT project

(Graphical Visual Instructional Design Languages

for Teachers) is based on this approach. Its main

objective is to investigate Model Driven Engineering

(MDE) techniques for supporting the speciﬁcation

of LMS-centered graphical instructional design

languages and the development of dedicated editors.

This paper deals with one central challenge: raising

the pedagogical expressiveness of LMSs learning

design semantics by using MDE techniques.

To this end, we detail in section 2 our research

context, including the presentation of the GraphiT

project, as well as a position of our current results.

Section 3 of this paper is dedicated to a survey and se-

ries of interviews we conducted with designers in or-

der to collect needs and requirements for the Moodle

LMS. A global presentation of our abstraction propo-

sition is done in section 4: a 4-level abstract syntax,

Loiseau E., Laforcade P. and El Mawas N..

Turning Recurrent Uses of E-learning Tools into Reusable Pedagogical Activities - A Meta-modeling Approach Applied to a Moodle Case-study.

DOI: 10.5220/0005434000640076

In Proceedings of the 7th International Conference on Computer Supported Education (CSEDU-2015), pages 64-76

ISBN: 978-989-758-107-6

 2015 SCITEPRESS (Science and Technology Publications, Lda.)

formalized as a metamodel, for the future learning de-

sign language. Section 5 focuses on the ﬁrst-level

and includes the proposition of a speciﬁc method to

identify the pedagogical activities and their bindings

to LMS tools. We also use the speciﬁc weaving lan-

guage we developed to formally capture these bind-

ings. Section 6 illustrates our propositions by a con-

crete learning scenario.

2 RESEARCH CONTEXT

2.1 LMS and Instructional Design

LMS development is usually based upon an educa-

tive theory rationale, or some speciﬁc pedagogical

approach. For example, Moodle claims a socio-

constructivist pedagogy philosophy (Dougiamas &

Taylor, 03). Widespread LMS generally follow such

an orientation because of the various production

and communication tools provided. LMSs are the

activity-centered evolution of former learning-object-

centered TEL-systems. Indeed, current LMSs pro-

vide designers with some numerous functionalities

that can be used to realise various learning activities

and are not restricted to providing resource access to

students.

Nevertheless, activity-centered standards like the

de facto IMS-LD (IMS, 03) fail to integrate exist-

ing LMSs. Experiments on extending Moodle to im-

port IMS-LD learning scenarios proved that adapt-

ing existing LMSs requires some complex and heavy

re-engineering (in particular integrating a dedicated

runtime-engine) in order to overcome the limitations

of the platform features and semantics (Burgos et al.,

07). Educational Modeling Languages (Berggren et

al., 05) fail to provide a support for operationaliz-

ing EML-conformed learning scenarios into existing

LMSs. For now widely spread LMSs like Moodle still

do not propose an IMS-LD compliance.

Moodle proposes its own format for importing

questions into quizzes. Our idea is to generalise it

to the whole instructional design aspects. Similarly

to the SCORM (ADL, 14) compliance about Learn-

ing Objects, the rationale of the GraphiT project is

based on the idea that LMSs should make explicit

their learning design format in order to facilitate the

import/export of compliant learning scenarios, and, in

addition, to encourage LMS-dedicated instructional

design editors.

2.2 Existing and Relevant Learning

Design Tools and Languages

The project main goal is to study the possibilities and

limitations about the pedagogical expressiveness of

operationalizable languages. According to the clas-

siﬁcation of Educational Modeling Languages pro-

posed by (Botturi et al., 06), our objectives map to

formal languages, i.e. with closed set of concepts

and rules for composing the designs, with an imple-

mentation level of elaboration, i.e. the highest level

of detail achieving maximum precision, with a visual

notation system.

The Glue! architecture, including the Glue!PS

editor (Alario-Hoyos et al., 12), and the CADMOS

editor (Katsamani et al., 12) are two recent research

works sharing our learning design criteria about ped-

agogical sound and executable learning design edi-

tors. They both propose an LMS-independent solu-

tion offering an LMS deployment feature towards the

most widespread and used platform: Moodle (Moo-

dle, 14). They both achieve the deployment by gen-

erating a Moodle course backup with all the infor-

mation and mapping their own data model concepts

to Moodle data model concepts; this backup is then

imported and deployed within a Moodle course us-

ing the Moodle restoration process. Such approaches

result in semantics adaptations and semantics losses

during their internal mappings because of the gap be-

tween the instructional design language, the speciﬁc

learning design capabilities as well as features of the

targeted LMS.

Other research (Abdallah et al., 08) shows that

model transformations techniques from the MDE the-

ories and tools can be useful to translate a designer-

centered and LMS-independent learning scenario to

a one that is LMS speciﬁc. Nevertheless, they

also highlighted the complex transformation model to

specify, the LMS metamodel to capture, the seman-

tics losses during translation, and the requirement of

an LMS-dedicated tool for embedding the scenarios

into the LMS.

2.3 Overview of the Graphit Project

from an MDE Perspective

The project methodology consists in exploring how

Model Driven Engineering and more particularly Do-

main Speciﬁc Modeling techniques and tools can be

relevant and useful to help in developing learning de-

sign editors that (1) are focusing on learning design

for a speciﬁc existing LMS, (2) are enough expres-

sive for abstracting the LMS’s implicit learning de-

sign, and (3) are machine-readable, or executable, to

TurningRecurrentUsesofE-learningToolsintoReusablePedagogicalActivities-AMeta-modelingApproachAppliedto

aMoodleCase-study

be fully traduce into ﬁrst LMS implementation set-

tings.

Brieﬂy, MDE is a software development method-

ology which focuses on creating and exploiting for-

mal domain models and meta-models, rather than on

producing code. MDE is also a large research ﬁeld

about specifying / executing / transforming / com-

posing (meta-)models. It comes with many speciﬁc

tools to support all these activities. DSM can be con-

sidered as a speciﬁc MDE process. It involves the

systematic use of domain-speciﬁc languages. These

languages tend to support formal higher-level abstrac-

tions in contrast to semi-formal general-purpose mod-

eling languages like UML.

The approach and architecture we propose is dif-

ferent from other existing approaches. We propose an

LMS-dependent architecture that only focuses on one

existing LMS in order to provide an instructional de-

sign language that will be speciﬁed and tooled. Our

idea is to conduct the platform abstraction in accor-

dance with the formalisation of future learning sce-

narios. We do not aim at extending the LMS se-

mantics with new add-ons/plugins, enriching it with

more pedagogical-oriented features. Our objective is

to support learning scenarios speciﬁcation in confor-

mance with the LMS semantics (its abilities as well as

its limitations). Furthermore, we do not aim at only

providing a notation layer on top of the LMS meta-

model. The results of past experiments (Loiseau &

Laforcade, 13) show that the best solution (expres-

siveness/LMS compliance ratio) relies on extending

the LMS metamodel. However, it requires a strong

metamodeling expertise to reduce the developing cost

while restoring the LMS compliance. This solution

also highlights the importance to drive the expres-

siveness (and semantics) extension of the initial meta-

model with the binding capacity. This paper focuses

on our further results and propositions about this is-

sue.

By extending the LMS metamodel we also ex-

tend the abstract syntax of the instructional design

language and thus lose the LMS-compliance format.

We plan to restore it by DSM (Domain-Speciﬁc Mod-

eling) techniques (weaving and transformation mod-

els). We aim at guaranteeing that learning scenar-

ios could be fully operationalized into the LMS with-

out semantics losses. Our approach can take advan-

tage of this LMS-dependance but it has also the dis-

advantage to be restricted to one LMS and one of

its versions. The LMS instructional design seman-

tics has ﬁrst to be identiﬁed and formalized as a do-

main metamodel. This metamodel drives the elabo-

ration of an XSD (XML Schema Deﬁnition) schema

that will be used as a format reference for the API

to develop. This API will be used through an import

facility available to teachers-designers in their LMS

courses. Its function is to parse the XML-based sce-

nario and ﬁll-up the LMS database. According to

DSM techniques and tools (like the EMF/GMF ones

for example (Eclipse, 14)), the visual instructional de-

sign language will be composed of an abstract syn-

tax (based on an extension of the LMS metamodel)

from which the graphical, tooling and mapping mod-

els will be derived. The editor will also be devel-

oped using the code-generation feature of DSM tools.

The produced scenarios have to be compliant with the

initial LMS meta-model to be deployed by the API.

We propose then to run two kinds of model transfor-

mations. The ﬁrst one will consist of various, ﬁne-

grained transformations that will be during design-

time: it will show some LMS mappings to teacher-

designers in order to help and guide them in the de-

sign process. The second transformation, unique and

large, will be used as an export feature (after design-

time).

The main challenge of this project is to cre-

ate enough abstraction from the LMS instructional

design semantics to provide teachers with some

pedagogically-sound higher design building blocks.

The LMS expressiveness and limitations have to be

overcome in order to offer teachers some instructional

design mechanisms closer to their practices and needs

about specifying and sequencing learning activities.

Although the GraphiT project deals with differ-

ent LMSs for guaranteeing the reproducibility of its

results, we focus primarly on the Moodle platform

which is the most popular open-source LMS.

3 COLLECTING

TEACHERS-DESIGNERS’

REQUIREMENTS

We evaluated several theoretical sources (Conole et

al., 04) as well as made practical exchanges with ped-

agogical engineerers in order to sketch our proposi-

tion orientations. Following, we decided to conduct a

larger survey with complementary interviews to ver-

ify our initial assumptions, to collect feeback about

our project orientations and positions, and to identify

more precisely end-users’ practices, needs and learn-

ing design tools requirements about the Moodle LMS.

3.1 Global Overview of the Survey

We conducted an online survey that was diffused

through international French-speaking higher educa-

CSEDU2015-7thInternationalConferenceonComputerSupportedEducation

tion institutions during a 4-week period. This sur-

vey addressed teachers and pedagogical engineers us-

ing existing LMSs. The survey was composed of 21

mandatory questions, most of them accepting mul-

tiple answers. Some questions were conditioned to

the selection of previous speciﬁc answers. For exam-

ple, the ﬁrst 8 questions (relative to the global design

of courses) are LMS-independent, whereas the other

ones were only available to people using Moodle (the

LMS we wanted to focus on). We received and anal-

ysed 208 results. We only sketch here the most no-

ticeable and relevant points in relation to the focus of

this paper.

74% of those polled use an LMS in addition to

their face-to-face courses (32% only for this pur-

pose), 52% for distant courses and 37% during the

face-to-face sessions. Main uses of the LMS con-

cern the document transmission (91%), collection

of works (52%), support for collaborative activities

(47%), evaluations (47%), and new pedagogical prac-

tices (58%). On average, half of those polled consid-

ered having explored the LMS alone. Those who did

not consider themselves as novices (56%) stated that

they had improved their LMS knowledge on their own

at 73%.

Although half of Moodle users consider that the

global user-interface of a course is easily understand-

able, only 33% consider that the form-oriented pa-

rameterization screens are understandable. From a

learning design perspective, they sketch all (38%) or

part (37%) of the learning scenario before setting-

up the equivalent course upon Moodle. 43% of

this sub-population have met some difﬁculties dur-

ing this manual step and have felt constrained in

adapting their initial scenarios and intentions (12%

failed to adapt the scenario). A majority of Moo-

dle designers use the basic functionalities like the

move left/right (64%), the hide/show (84%) param-

eters. Half of the group graded students’ produc-

tions and use Moodle’s groups and groupings when

required. 62% used the restrict access settings but

only 34% the activity completion. 15 of 22 Moo-

dle standard activities/functionalities (note that insti-

tutions can add/remove these blocks) are not well

known by an average of 50% (sometimes more) of an-

swerers, whereas the 7 others are regularly used. The

Forum is largely preferred to the Chat feature to fos-

ter communication. For the setting-up of exercices,

Assignment (47%) and Quiz (37%) are preferred to

Hot Potatoes (15%) or Lesson (19%). The Wiki is the

most preferred collaborative tool (23%) among others

(Journal 8%, Workshop 8%).

3.2 Most Relevant Points from

Interviews Analysis

From most relevant answerers that agreed to be con-

tacted we conducted 20 one-to-one interviews, mostly

by distant devices. Interviewees were selected de-

pending on their instructional design expertise about

the Moodle platform.

They agree that Moodle is useful for simple ped-

agogical objectives but is time-consuming for elab-

orating more complex learning situations. Settings

screens are considered too complex and difﬁcult to

handle. These screens mix pedagogical and technical

parameters, requiring to test and observe the pedagog-

ical implications of all combinations. Some intervie-

wees stated that they encourage to use default param-

eters and then, hinder the setting-up of more complex

activities.

A majority of interviewees accept the idea of both

an external learning design editor dedicated to Moo-

dle and an import block available through the Moo-

dle internal design space to automatically set-up the

course (the external feature allowing ofﬂine designs

and the graphical notation helping to visualize the

scenario at design-time). They approve the approach,

emphasising its relevance if templates or concrete

cases about pedagogical uses of Moodle tools can be

handled within the editor. They highlight the need

for a language/editor covering large pedagogical uses

but without being too generic. Some of them consider

important to continue using the editor for adapting the

scenario after the import step although they agree that

a round-trip use of both editor and Moodle can be an

obstacle.

One issue highlighted is that practitioners did

not expressed common design practises, as we ex-

pected them to, mainly because of the heterogene-

ity of their Moodle expertises and pedagogical back-

grounds. Nevertheless they have in common to think

about Moodle tools according to their basic pedagog-

ical uses. Indeed, they all point the heavy parameteri-

zations of tools and resources and the need for having

an abstract view of what are the pedagogical uses in

order to help and guide them in selecting and conﬁg-

uring the right implementation activities.

3.3 Requirements for Our

Language/editor

From all these practitioners feedback we listed some

speciﬁc requirements for our Moodle language/editor

to develop. First, they mentioned the need for the

graphical authoring-tool to allow designers to select

pedagogical blocks on top of the LMS semantics as

TurningRecurrentUsesofE-learningToolsintoReusablePedagogicalActivities-AMeta-modelingApproachAppliedto

aMoodleCase-study

well as with Moodle building blocks to compose with.

In their mind, the editor will not have to strictly

follow a top-down process from abstracted speciﬁ-

cation elements to implementation one expressed in

terms of Moodle; abstractions from Moodle and its

own concepts should be mixed up together according

to practitioners’ expertise about instructional design

(speciﬁcation and implementation concepts mix).

Secondly, they are interesting in the idea that map-

pings from pedagogical design blocks to Moodle con-

cepts can be showed to practitioners (default map-

ping) and adapted if required (mapping adaptation).

This design approach could help practitioners in the

appropriation of the pedagogical constructs and guide

them in designing more abstract learning scenario

while mastering the translations into LMS elements.

Another design point highlighted (declarative

non-visible information) is about the possibility to

design and declare within the learning scenario some

information that do not required to be mapped into

LMS concepts or just mentioned as non-visible labels

(for students/tutors) for the teacher him-self: infor-

mation about the face-to-face sessions mixed up with

the LMS-centered ones, about pedagogical strategies

or pedagogical objectives, about activities to realize

on the LMS at a speciﬁc runtime moment accord-

ing to concrete data (enrolled students, dates, etc.).

Finally, another design need was to help teachers

in sequencing the course in more advanced struc-

tures (choices, sequences) with elements showed one-

by-one according to their progress (advanced activ-

ity structures). Indeed, these can be done manu-

ally but it requires to parameterize many low-levels

and technical-oriented properties (achievements, re-

stricted access conditions...) that they would appreci-

ate not to have to set up by themselves.

4 ABSTRACTING THE LMS

METAMODEL

We plan to study how the LMS semantics about in-

structional design could be abstracted from two per-

spectives: a theoritical one, generalizable to different

LMSs, and a practical one about the special case of

the Moodle LMS. We chose to follow a bottom-up

approach by focusing at ﬁrst on the Moodle LMS. Ac-

cording to practitioners’ needs, the abstraction could

consist in raising the recurrent LMS uses support-

ing some learners and/or teachers activities. From an

activity theory perspective(Engestrom, 87)(Benson et

al., 08), such activities should involve LMS’s bind-

ings of subject, objects, tools/artefacts, community,

division of labor and rules. Because our survey and

interviews highlight a special need to ease the param-

eterization of Moodle tools and resources for setting-

up activities, we decided to focus at ﬁrst on raising

these tools and resource semantics, and to study later

the other aspects.

The following sections present these abstractions

in relation with their formalizations for the Moodle

LMS. The metamodel from Figure 1 can be consid-

ered as part of the general abstract syntax of the in-

structional design language to be developed. This

part focuses on the abstraction of Moodle tools and

resources.

4.1 Tool-or-resource-based Pedagogical

Activities

We deﬁne an LMS-abstract pedagogical activity as an

encapsulation of parameters a teacher has to set-up

when using a tool (or resource) for a speciﬁc peda-

gogical use. From a single tool, for example a forum,

a teacher can design several pedagogical use, depend-

ing on its conﬁguration: to provide news to students,

to set up group work, to propose a peer reviewing ac-

tivity, etc.

Because several LMS functionalities can be used

for the same pedagogical purpose, we have to ﬁnd the

discriminatory criteria that can guide to identify the

right tool and default mapping (as well as the rela-

tions to objectives, resources, groupings, etc. that are

involved in the right setting-up of the pedagogical ac-

tivities).

To be used appropriately, this ﬁrst abstract block

requires a name, a description, and speciﬁc proper-

ties (the former discriminants), set at design-time by

practitioners, that will drive the default mapping. For

example an exchange activity, involving student com-

munication, could either rely on a forum or a chat,

depending on a synchronous property. The mappings

will not be limited to the parameterization of a tool. It

will also impact some other elements in relation with

the tool/resource: grades, objectives, groupings, re-

striction access and achievements rules, etc.

4.2 Activity Structures

According to (Gedera & Williams, 13), successful im-

plementation of an online course that is facilitated by

an LMS needs careful planning, including structural

strategies in the design of the course. In order to

ease and assist the practitioners when assembling and

setting-up combinations of activities or resources we

propose then usual structural elements (selection, se-

quence, conditional activities, etc.). These blocks will

be composed of activities or other activity structures.

CSEDU2015-7thInternationalConferenceonComputerSupportedEducation

Figure 1: The abstract syntax of an instructional design language on top of the Moodle metamodel.

Every instructional design language feature some of

them. In the case of Moodle they will be concretely

translated as complex combinations of labels (stating

the structure name, kind and use for users), shifted

content (move left/right feature) according to the ac-

tivity structure components in the learning scenario.

After various translations and mappings until reach-

ing the LMS low-level elements, all its content will

be parameterized (restrict access, visibility, achieve-

ment...) with appropriate properties in order to set up

the desired behavior.

4.3 A 4-levels Abstract Syntax

The global architecture we propose for the abstract

syntax of the Moodle-centered instructional design

language is composed of four levels. Figure 1 illus-

trates our proposition with a graphical representation

of the ecore domain model.

Level 1 ﬁts the Moodle metamodel. Readers have

to consider Figure 1 as a part of the whole metamodel.

Only important structural relations and concepts are

depicted because of our current interest. Level 1 el-

ements (restricted to the Moodle activities the Moo-

dle name given to the tools - and resources) can be

directly used by teachers-designers and parameter-

ized for building a learning session. From the Moo-

dle metamodel point of view these elements require a

global Course and a Section container to be attached

to. In the extended metamodel they will be speciﬁed

at ﬁrst as child of level 4 elements. The model trans-

formation, at post-design-time, will deal with restor-

ing a model in full-compliance with the Moodle meta-

model: creation of the global Course instance, Section

instances, attachment of all the corresponding Moo-

dle elements according to the orders and positions de-

ductible from the source scenario.

Level 2 includes our pedagogical activities. They

are composed of Level 1 elements, i.e. Moodle activi-

ties and resources. Level 3 captures the activity struc-

tures. The activity structures are composed of Level

3 elements speciﬁed during the design-time. Finally,

the fourth level is the contextual level focusing on the

global structure of the learning session in relation to

the different face-to-face, complementary, distant ses-

sions or other teacher-deﬁned customized sessions.

Such Level 4 elements rely on the Moodle section

concept. Indeed, Moodle only proposes sections into

the space of the course for aggregating the tools and

resources. However, designers have at their disposal

an indentation feature (position property in the Moo-

dle metamodel) to shift activities and resources in or-

der to visually indicate their collective relationships.

This position property will be used by the dynamical

mappings, in order to position the corresponding ele-

ments in accordance to the source element position in

the global learning scenario.

The relations with a red composition indicate that

the content will not be displayed in the future con-

crete syntax (notation) as nested elements but will

appear in another sub-diagram where the parent con-

tainer will be the root canvas. Differently, the green

composition indicates that content will be showed as

nested elements of the parent container in the same di-

agram. Finally, the nextE reﬂexive relation allows, by

inheritance, to provide a previous/next information to

TurningRecurrentUsesofE-learningToolsintoReusablePedagogicalActivities-AMeta-modelingApproachAppliedto

aMoodleCase-study

sequence the various elements within their dynamic

pedagogical context (the ordering concerns the child

elements sharing a same Level Element parent).

The leaf meta-classes from ﬁgure 1 (dark ele-

ments) sketch some examples of future elements.

They are on purpose not showing their attributes (for

ease of reading). However each of them owns speciﬁc

properties in accordance with the different in-progress

formal speciﬁcations we are studying about the Moo-

dle instructional design semantics, pedagogical activ-

ities, and activity structures.

4.3.1 Overview of the Proposed Learning Design

Tool

The proposed authoring-tool will directly propose to

practitioners the level-4 elements in the tool palette.

Indeed, these elements are required to map to Moodle

sections in order to sequentially structure the course

skeleton. Sessions that do not rely on Moodle features

can also be described if designers need an overall

view of a global module/course larger than the ones

involving the use of an LMS. Other level-4 elements

will then open an empty sub-diagram when double-

clicked. It can then be used to arrange levels 3-to-1

elements from the new palette. Indeed, practitioners

can then choose the method (top-bottom, bottom-up),

the description level (speciﬁcation versus implemen-

tation) and the elements to select, combine and adapt.

Pedagogical activities can be opened up as another

sub-diagram containing the default mapping to levels

1 elements. Every mapping can be adapted and mod-

iﬁed by deleting/adding new elements (according to

those accepted under the parent element) or modify-

ing the elements properties. This layer-oriented nota-

tion and functionalities ﬁt the practitioners’ need de-

picted in section 3.3.

5 FOCUS ON THE

PEDAGOGICAL ACTIVITIES

5.1 An Identiﬁcation Method

In order to identify the most appropriate tool for a

speciﬁc pedagogical activity, we followed these three

steps: (1) analysis for each Moodle tools of its re-

current uses (bottom-up method), (2) identiﬁcation

of tools offering common uses (top-down method),

and (3) speciﬁcation of discriminating criteria to drive

the selection of a suitable tool. Moodle 2.4 offers

7 resources (Book, Page, Label, IMS content pack-

age, File, Folder, and URL) and 13 activities (Fo-

rum, Database, Glossary, Assignment, Lesson, Quiz,

Workshop, SCORM package, External tool, Choice,

Survey, Wiki, and Feedback). We have study the

recurrent uses of these Moodle’s activities and re-

sources. We notice that some activities/resources can

be diverted to serve for different uses. For exam-

ple, everyone knows that the Forum is used for dis-

cussion reasons but it can also be used to allow stu-

dents to introduce themselves in a course or to con-

sult a Frequently Asked Questions (FAQ) or to share

documents between learners. After looking at all

Moodle’s activities/resources uses, we have identiﬁed

those supporting the same uses. Three tools can be

used to consult a FAQ: the Forum, the Wiki, and the

Glossary.

We have then speciﬁed discriminating criteria to

help a teacher in deciding which tool he must use if

he has many choices. We chose m×n matrix A format

to present these discriminating criteria (A has m rows

and n columns, ﬁrst row and column are headers and

not part of the matrix data) according to seven rules :

R1 The pedagogical activity name is only from a

teacher perspective if no students are concerned

(= with parameter hide on). For example, for a

survey, we choose the expression answer a sur-

vey (students viewpoint) instead of create a sur-

vey (teachers viewpoint). Note that A

presents

this pedagogical activity.

R2 Tools participating to the realization of the activ-

ity are the elements A

...A

R3 Discriminating criteria are the elements

...A

R4 Discriminating criteria are expressed as much as

possible as a pedagogical question designers have

to answer by Yes or No.

R5 Cells intersecting a discriminating criterion and a

tool must embed all answers that can implied to

choose this tool (Yes/No are both possible if this

criterion is not directly discriminant for this tool,

i.e. the tool can support both pedagogical cases).

R6 A valid discriminating criterion must cause at

least one different answers for one tool.

R7 The matrix is terminated if there is no similar

combination of answers for two tools.

An unachieved matrix indicates to experts that

they have to add one more discriminating criteria and

verify again the rule R7.

Table 1 shows an example of identiﬁcation ma-

trix for the pedagogical activity (PA) ”Answer a poll”.

Four Tools can support this activity: Quiz (T1),

Choice (T2), Feedback (T3), and Survey (T4). Ex-

perts have found 7 discriminating criteria. Each crite-

rion is presented in the form of a question:

CSEDU2015-7thInternationalConferenceonComputerSupportedEducation

- (C1) More than one question?

- (C2) Only multiple choice questions?

- (C3) Pre-populated with questions?

- (C4) Time limit?

- (C5) Anonymous?

- (C6) Graded?

- (C7) Feedback after submission?

Table 1: Example of identiﬁcation matrix.

T1 T2

Yes/No No Yes/No Yes

No No No Yes

C4 Yes/No No No No

C5 No No Yes/No No

C6 Yes No No No

C7 Yes No Yes No

In table 1, we have three different answers that

can imply these four tools: Yes, No, and Yes/No. For

example with a survey (T4), we can have more than

one question (A

= Yes),only multiple choice ques-

tions are allowed (A

= Yes), it is a pre-populated

survey with questions (A

= Yes), it can not have a

time limit/countdown timer for students’ navigation

= No), it is always nominative (A

= No), it can

not be graded (A

= No), and students can not have

a feedback after their submissions (A

= No). Note

that a designer can reply to C1, C2, C3, C4 , C5, C6,

and C7 in any order. Some combinations cannot lead

to a speciﬁc tool choice for two reasons: (1) a non-

valid combination, or (2) a non-response to all ques-

tions. In the ﬁrst case, the experts will be notiﬁed to

adapt their pedagogical choices while in the second

case they will be asked to precise more choices.

Such identiﬁcation matrix has to be completed by

additional information in order to precise the general

(whatever the answers that guide the binding like a

Tool’s name) or contextualized (depending on some

speciﬁc answers like a tool’s format) parameters for

the related LMS activity or resource.

5.2 A Weaving Language to Formalize

the Mappings

According to our Model Driven Engineering research

framework, we can use model transformations to

achieve the mappings speciﬁed by experts. The trans-

formations will be run at design-time, to add mapped

elements to the model and populate the sub-diagrams.

Such transformations are complex (proportionally to

the mapping complexity) and numerous, thus costly

to write.

We on purpose propose to use the model weav-

ing technique we studied in (Loiseau et al., 14) to

capture the mapping semantics in dedicated weaving

models and automatically generate models transfor-

mations. From a practical viewpoint, thanks to the

matrice and additional information from an LMS ex-

pert using our method and formalisms depicted in sec-

tion 5.1, an engineer will formalize the mappings in

a weaving model, using a tree based editor. He can

then run a generic High Order Transformation (HOT)

that will generate the concrete ”mapping transforma-

tions”. These ﬁnal transformations can then be inte-

grated within the graphical editor to be run at design-

time.

The weaving models can be expressed using a

weaving language, based on a generic weaving meta-

model we designed. This weaving metamodel deﬁnes

the ”syntax” of the mapping/weaving model. Each

mapping (or binding) has one source element and one

or several targets (chosen from the extended instruc-

tional design metamodel). Targets can have condi-

tions on whether they have to be instantiated or not,

attributes can be set to speciﬁc values (also with con-

ditions). Figure 2 is a screenshot of the weaving

editor. It is used to formally specify as a weaving

model the corresponding binding speciﬁed by LMS’s

experts. Concretely, the left part of the ﬁgure should

concern level-2 elements of our instructional design

metamodel for Moodle (1) whereas the right part

should only concern level-1 elements, i.e. Moodle el-

ements. Because our instructional design metamodel

includes the Moodle one, source and target metamod-

els involved in the weaving are the same. The central

part of the ﬁgure is the concrete tree-based editor for

specifying bindings.

Figure 3 is an example of weaving model related

to the binding example from the table 1. It speciﬁes

the corresponding translations for all possible combi-

nations by using AND/OR/! operators. Such weaving

model is realized by following the matrice formalism,

tool by tool. Informations about the tools parameteri-

zations are deduced from the additional informations

given by experts.

We used languages and tools from the Epsilon

(Epsilon, 09) project to build a software framework

fulﬁlling our model weaving requirements. This

project is compliant with the Ecore formalization of

metamodels we already used to formalize the vari-

ous metamodels we illustrated. This Ecore format

is from the Eclipse Modeling Framework (Eclipse,

14). Weaving models are edited through ModeLink,

a three pane editor displaying the source and target

metamodels in side panels (which are the same in

our use case). The ﬁnal ”mapping” transformations

TurningRecurrentUsesofE-learningToolsintoReusablePedagogicalActivities-AMeta-modelingApproachAppliedto

aMoodleCase-study

Figure 2: Screen caption of our weaving tool for formalizing the bindings.

Figure 3: Example of resulting weaving model (tree-based

representation of a concrete XML ﬁle).

are expressed using Epsilon Object Language (EOL),

and are generated through a Model-to-text transfor-

mation using EGL language. This last transformation

replaces the HOT traditionally used in model weaving

environments.

6 A Learning Scenario Example

We on purpose propose to illustrate our proposal by

formalizing a very simple but representative learning

scenario for the Moodle LMS. We propose at ﬁrst a

brief textual description, then the equivalent speciﬁ-

cation as a model conformed to the dedicated meta-

model we proposed in section 4 (Figure 4 is a screen-

shot of the EMF-tree-based model editor).

6.1 Scenario Description and

Formalization

The learning scenario is composed of two learning

sessions. The ﬁrst one is a lecture session for which

the teacher only want to provide learners with a Re-

CSEDU2015-7thInternationalConferenceonComputerSupportedEducation

Figure 4: Example of simpliﬁed learning scenario com-

posed of elements from the 4 levels.

source consultation corresponding to his face-to-face

course material. This pedagogical activity has the

quantity property set to ”one” and the location set to

”local”. These properties will lead the dynamic map-

ping process to add the File Moodle element to the

scenario.

Figure 5: Same example of simpliﬁed learning scenario

with generated contextualized tools and resources bindings.

The second learning session is a practical work

that the teacher wants to realize in face-to-face within

a computerized classroom. He would like to use the

Moodle platform for supporting a sequence activity

structure embedding 4 sub-components. The ﬁrst one

is another Resource consultation. This time, the prop-

erties set to ”many” (quantity) and ”local” (location)

by the teacher will lead the transformation process to

add a Folder tool. The second sub-element is a Brain-

storming pedagogical activity. Its orientation prop-

erty set to ”discussion” leads to propose a Forum tool.

Similarly the third one is another pedagogical activity

Report writing leading to a Wiki tool because of the

collaborative property set to ”true”. Finally the fourth

sub-component is a Guidance activity that aims at re-

minding the teacher to evaluate the synthesis in the

wiki. Thanks to a public property set to ”tutor” it

leads the mapping process to set the corresponding

Label to be invisible (visible=”false”) to students (it

will only be displayed to the teacher).

The teacher can change at any time the activities

properties, leading to other mapping adaptations. He

can also manually delete the mapping elements, re-

arrange their order, or add some other elements. Fig-

ure 4 shows a global overview of the learning sce-

nario elements including all the automatic mappings

according to the various properties and values (not de-

picted within the ﬁgure).

Figure 6: Example of weaving model specifying mappings

from Figure 4.

6.2 Current Tooling Prototype

We are currently working on the development of a

prototype adding a notation layer on top of the ab-

stract syntax we propose for the Moodle-centered in-

structional design language. We choose for now the

Sirius tooling (Sirius, 14) because it allows to quickly

deﬁne custom multiview for workbenches with less

technical knowledge compared to the well-known

GMF tooling (from (Eclipse, 14)). The notation, or

concrete syntax for our instructional design language,

is derived from the abstract syntax formalized as an

Ecore metamodel. Sirius also facilitates the develop-

ment of dedicated graphical tools by generating most

of features (diagrams, trees, tables, etc.) from the

sirius-speciﬁc model we build when using it. It re-

duces the cost and complexity of developing a graph-

ical editors.

We succedeed in integrating the mappings trans-

formations within this prototype. For now, when

a user double-clicks on a session within the ﬁrst

TurningRecurrentUsesofE-learningToolsintoReusablePedagogicalActivities-AMeta-modelingApproachAppliedto

aMoodleCase-study

Figure 7: Different screen captions of our current prototype.

level diagram (not depicted in Figure 7), it opens a

new diagram where he can mix elements from lev-

els 1 to 3 according to his Moodle expertise (mid-

dle part). Pedagogical properties of level-2 elements

can be set at this stage. When these level-2 ele-

ments are double-clicked, a transformation process

is launched for checking all transformation rules au-

tomatically generated from the weaving models we

produced. The execution of an eligible rule modiﬁes

at run-time the current scenario by adding the corre-

sponding binding towards Moodle elements (level-1).

Bottom part of Figure 7 shows resulting mappings.

The result is part of the pedagogical scenario: it can

be modiﬁed by adding/editing/removing new level-1

elements. Mappings can also be updated if a user

changes the pedagogical properties of a level-2 ele-

ment.

7 CONCLUSIONS

7.1 A First Tools-centered Abstraction

This paper proposes a speciﬁc LMS-centered ap-

proach for raising the pedagogical expressiveness of

its implicit learning design semantics. We discussed

how the LMS low-level parameterizations could be

abstracted in order to build higher-level building

blocks capturing some recurrent resources or tools

uses into pedagogical activities. We also presented

a speciﬁc method for helping and guiding LMS ex-

perts to describe how these activities should be binded

to appropriated tools or resources. In addition we

propose a speciﬁc model weaving approach for for-

malizing these mappings. These resulting weaving

models will drive at run-time and in real-time the au-

tomatic translations when using the authoring-tool.

Based on a Moodle application, we present and illus-

trate our approach by formalising the abstract syntax

of a Moodle-dedicated instructional design language

following a speciﬁc 4-levels architecture. Thanks to

illustrative examples and an overview of our current

prototyping editor, we concretely argued and veriﬁed

our propositions.

7.2 Ongoing Work

The current abstract syntax proposition still has to be

improved in order to allow the declaration of didac-

tic objectives to the various Level 4-to-1 elements.

According to the objective status (goals/competencies

for teacher or activities objectives for learners under-

standing), the mapping could lead to Moodle Out-

comes, attached to the root Course and referenced by

the direct or indirect corresponding Level 1 elements,

describing labels, or simply a description ﬁeld ﬁlling-

up. Similarly, roles or groups have to be included in

order to allow the division of labour in the learning

scenario. Mappings to the Moodle concepts of Group

and Grouping will be studied.

Also, in our approach the 4-levels extended meta-

model will not allow to serialize future learning sce-

narios in conformance with the LMS format (source

metamodel): a global transformation is required to re-

store this conformance. This transformation will be

available as an export feature from our authoring-tool.

CSEDU2015-7thInternationalConferenceonComputerSupportedEducation

We are planning an experimentation of the proto-

type with end-users (teachers), once it has reached a

stable enough version. The main objective will con-

cern the validation of the language, along with the de-

sign approach, more than the tool itself (in terms of

ergonomy or user experience for example). By pro-

viding teachers with only general guidelines about

the scenario to produce, we would like to evaluate

how they are able to design a pedagogically sound

course with the provided editor. We are interested in

observing which abstraction levels and elements they

will use, if they are following a top-down or bottom-

up approach, if they will make use of the default-

mapping feature, and other speciﬁc contributions we

are proposing.

ACKNOWLEDGMENTS

This article is part of the GraphiT project, a 42-

months funded project of the French research agency.

REFERENCES

Abdallah, F., Toffolon, C., Warin, B. (2008). Models trans-

formation to implement a Project-Based Collabora-

tive Learning (PBCL) Scenario : Moodle case study.

The 8th IEEE International Conference on Advanced

Learning Technologies, pp. 639–643. IEEE Computer

Society, Washington DC, USA.

Abedmouleh A., Oubahssi L., Laforcade P., Choquet C.

(2008). An analysis process for identifying and for-

malizing LMS instructional language. The 7th Inter-

national Conference on Software Paradigm Trends,

pp. 218–223. ScitePress.

Advanced Distributed Learning (2014). The

SCORM speciﬁcation. Retrieved from

http://www.adlnet.gov/scorm/

Alario-Hoyos, C., Munoz-Cristobal, J.A., Prieto-Santos,

L.P., Bote-Lorenzo, M.L., Asensio-Perez, J.I.,

Gomez-Sanchez, E., Vega-Gorgojo, G., Dimitriadis,

Y. (2012). GLUE!-PS: An approach to deploy non-

trivial collaborative learning situations that require the

integration of external tools in VLEs. The 1st Moodle

Research Conference, pp.77-85. Greece.

Benson, A., Lawler, C., Whitworth, A. (2008). Rules, roles

and tools: Activity theory and the comparative study

of e-learning. British Journal of Educational Technol-

ogy, Vol. 39, No. 3., pp. 456-467.

Berggren, A., Burgos, D., Fontana, J.M., Hinkelman, D.,

Hung, V., Hursh, A., Tielemans, G. (2005). Practical

and Pedagogical Issues for Teacher Adoption of IMS

Learning Design Standards in Moodle LMS. Journal

of Interactive Media in Education, special issue: Ad-

vances in Learning Design.

Botturi, L., Derntl, M., Boot, E., Figl, K. (2006). A

Classiﬁcation Framework for Educational Modeling

Languages in Instructional Design. Proceedings of

The 6th IEEE International Conference on Advanced

Learning Technologies, pp.1216-1220.

Burgos, D., Tattersall, C., Dougiamas M., Vogten, H.,

Koper, R. (2007). A First Step Mapping IMS Learning

Design and Moodle. Journal of universal computer

science (13). pp.924-931.

Conole, G., Dyke, M., Oliver, M., Seale, J. (2004). Map-

ping pedagogy and tools for effective learning design.

Computers & Education (4). pp.17-33.

Dougiamas, M., Taylor, P. (2003). Moodle: Using Learn-

ing Communities to Create an Open Source Course

Management System. The World Conference on Edu-

cational Multimedia, Hypermedia and Telecommuni-

cations, pp. 171178. Waynesville.

Eclipse Modeling Project Ofﬁcial Website. Retrieved from

2014. http://www.eclipse.org/modeling/.

Paige, R. F., Kolovos, D. S., Rose L. M. , Drivalos N.

, Polack F. A. C. (2009). The Design of a Con-

ceptual Framework and Technical Infrastructure for

Model Management Language Engineering. In Pro-

ceedings of the 2009 14th IEEE International Confer-

ence on Engineering of Complex Computer Systems

(ICECCS ’09), pp. 162-171. IEEE Computer Society,

Washington, DC, USA.

Engestrom Y. (1987). Learning by Expanding: An Activ-

ity Theoretical Approach to Developmental Research.

Helsinki: Orienta-Konsultit Oy.

Garrisson, D.R. & Kanuka, H. (2004). Blended learning:

Uncovering its transformative potential in higher ed-

ucation. The Internet and Higher Education (7). pp.

95-105.

Gedera, D. S. P, Williams, P. J. (2013). Using Activity The-

ory to understand contradictions in an online univer-

sity course facilitated by Moodle. International Jour-

nal of Information Technology and Computer Science

(IJITCS), 10(1), pp. 32-40.

IMS Learning Design speciﬁcation (2003). Retrieved from

http://www.imsglobal.org/learningdesign/index.html

Katsamani, M., Retalis, S., Boloudakis, M. (2012). Design-

ing a Moodle course with the CADMOS learning de-

sign tool. Educational Media International (49). pp.

317-331.

Koper R., Manderveld J. (2004). Educational modelling

language: modelling reusable, interoperable, rich and

personalised units of learning. British journal of edu-

cational technology, 35(5), pp.537-551

Loiseau, E. & Laforcade, P. (2013). Speciﬁcation of learn-

ing management system-centered graphical instruc-

tional design languages - A DSM experimentation

about the Moodle platform. The 8th International Joint

Conference on Software Technologies (ICSOFT13),

pp. 504-511. Scitepress.

Loiseau, E., Laforcade, P., Iksal, S. (2014). Model Weav-

ing and Pedagogy Mapping Abstraction Levels in

Instructional Design Languages. The 9th Interna-

tional Joint Conference on Software Technologies

(ICSOFT14), Vienna (Austria).

TurningRecurrentUsesofE-learningToolsintoReusablePedagogicalActivities-AMeta-modelingApproachAppliedto

aMoodleCase-study

Moodle Ofﬁcial Website. Retrieved from 2014.

https://moodle.org

Ormrod, J.E. (2011). Human Learning. Pearson College Di-

vision, Upper Saddle River.

Sirius Project. Retrieved from 2014.

http://eclipse.org/sirius/

CSEDU2015-7thInternationalConferenceonComputerSupportedEducation