Model of Articulation between aLDEAS Assistance Rules
Le Vinh Thai, Stéphanie Jean-Daubias, Marie Lefevre and Blandine Ginon
Université de Lyon, CNRS, Villeurbanne, France
Université Lyon 1, LIRIS, UMR5205, F-69622, Villeurbanne, France
1 INTRODUCTION
More and more applications are used in different
contexts: professional, personal and educational.
However, because of technical difficulties of
handling or use of these applications, their users can
abandon or under-exploit them and lose motivation
(Gapenne et al., 2002). In the educational context,
learners use various applications (pedagogical or
non-pedagogical applications) to acquire knowledge
(Ginon et al., 2014c). These applications must not
only resolve technical difficulties as applications
used in other interactive environments, but also give
learners the pedagogical feedback and guidance
which meet the pedagogical goals of teachers. For
example, a given hint as a pedagogical feedback
helps learners to do an exercise when they meet
difficulties. Or, a pedagogical guidance guides
learners to choose activities to work. However, these
pedagogical feedback and guidance are not always
supported by applications, especially non-
pedagogical ones used in the educational context.
Therefore, adding an assistance system is considered
as a solution for both technical and pedagogical
problems of an existing application. In details,
pedagogical assistance systems can meet both
technical (handling, use of applications) and
pedagogical assistance needs (hint, explanation,
guidance, etc.). However, the existing pedagogical
assistance is varied and complex. It can be a
complex pedagogical guidance to propose learning
activities suitable to learners (Antoniadis et al.
2004). For example, the remediation activities are
proposed depending on the progression of learners.
In a pedagogical activity, the pedagogical assistance
can have different modes to sequence of assistance
events (Melis et al., 2001), (Winke and MacGregor,
2001). Theses modes describe the articulation
between assistance elements. For instance, a
successive assistance gives one message after
another in order to guide learners. As part of my
thesis, we identified two research issues: “How to
help teachers to define the pedagogical guidance?”
and How to help teachers to define the articulation
between assistance elements?” In this paper, we
present our answer to second issue.
The AGATE project proposes the SEPIA system
(Ginon et al., 2014a) that allows assistance designers
(teachers) to add an assistance system in the existing
ILE (Interactive Learning Enviroment) by creating
and executing the aLDEAS rules (Ginon et al.,
2014b). SEPIA supports various types of
applications (windows, java, web and MacOS
applications), assistance techniques (textual, vocal,
enhancing, automatic actions, etc.) and it is the
independent of application domains. SEPIA is a full
solution to create rich assistance systems. However,
definition of the articulation between assistance
elements is still implicit and difficult. So, this paper
presents the evolution made to SEPIA to overcome
these limitations.
In this paper, first we present the AGATE
project and its major results: the SEPIA system and
the aLDEAS language. We also show that
expression of the articulation between assistance
elements of an assistance system in ILEs is a
complex task. Then, we present the different existing
modes of articulation through examples of assistance
(section 3.1). We too confront these modes to tools
that aim at the definition of assistance as well as
modes of articulation (section 3.2). These studies
allow us to propose a model of articulation between
aLDEAS assistance rules (section 4). Then, the
implementation of this model is presented (section
5). To validate our approach, we present some
results from our evaluation (section 6). Finally, we
give the general conclusions and the issues that
motivate the future works (section 7).
2 SEPIA SYSTEM
The AGATE (Approach for Genericity in Assistance
To complEx tasks) (AGATE, 2015) project aims at
proposing generic models and unified tools to enable
the setup of assistance systems in various existing
applications, that we call target-applications, by
applying a generic and epiphytic approach.
16
Thai, L., Jean-Daubias, S., Lefevre, M. and Ginon, B.
Model of Articulation between aLDEAS Assistance Rules.
In Doctoral Consortium (DCCSEDU 2016), pages 16-25
Epiphytic application is the application that is able
to perform actions in another application without
requiring any change to it. Thus, the functioning of
an epiphytic assistance system added in the target-
application doesn’t disturb the functioning of this
application (Paquette et al., 1996). The models and
tools proposed are specific neither to an application
nor to a domain. For that reason, we previously
proposed an adjunction process of epi-assistance
systems to a given target-application (Ginon et al.,
2014b). This process (Figure 1) consists of two
phases: the assistance specification and the
assistance execution in an epiphytic way.
The assistance specification is performed by an
expert of the target-application, called the assistance
designer. This preparatory phase enables the
designer to specify the assistance that he wishes for
a given target-application. The assistance execution
concerns end-users of the target-application. It is the
execution of the assistance designed by the designer;
it occurs at any use of the target-application by an
end-user. The epi-detectors make possible the
monitoring of the target-application. They detect the
events related to interactions between the user and
the target-application by exploiting the accessibility
libraries compatible with a type of applications
(windows, java, web applications…). Finally, the
epi-assistants handle the elaboration of the answer
to provide assistance to the end-user by pop-ups
windows, speech or animated agent as well as
highlighting a component on interfaces (for instance,
by colouring a component).
Figure 1: Adjunction process of epi-assistance systems
(Ginon et al., 2014b).
The aLDEAS language (a Language to Define
Epi-Assistance Systems) (Ginon et al., 2014b) is
proposed in order to connect the two phases of this
process. aLDEAS consists of three principal
elements: event wait (click on a button…),
consultation (of profile, of states of application,…),
assistance action (message, enhancing,…). This
language is completed by a rules pattern (Figure 2)
and also by other patterns facilitating the definition
of assistance actions (for example, step by step
pattern). A rule begins with event wait called
trigger event. When this event occurs, the launch of
assistance actions is immediate (upper path in
Figure 2), or is constrained by a condition (lower
path in Figure 2). This condition takes the form of a
consultation with the alternatives each associated
with one of these actions. Finally, the rule can be
terminated by end event that ends all elementary
actions launched by this rule. For example, the
example in Figure 2 shows a rule among many rules
which define an assistance system. This rule waits a
click on the button ‘help’ in order to verify the
answer of the learner and to provide an error
message when this answer is not correct (text written
by the learner is not equal to 1). This message is
closed after 10 secs.
aLDEAS and its patterns are implemented in the
SEPIA system (Ginon et al., 2014a) that consists of
two tools: an assistance editor and an assistance
engine. The assistance editor operationalizes the
assistance specification phase (upper part in Figure
1). It provides an interface that allows the assistance
designer to define an assistance system by creating a
set of aLDEAS assistance rules. The assistance
engine operationalizes the assistance execution
phase (lower part in Figure 1). It executes the
assistance system created in the previous phase by
executing its set of aLDEAS assistance rules.
Figure 2: aLDEAS rules pattern.
SEPIA and aLDEAS allow creating useful
assistance systems in various domains, among which
ILE. However, the creation of assistance systems
which can be found in ILEs is complex. Thus, the
principal objective of target-applications in ILE is
learning. Assistance systems for these applications
must be effective, suited to learners, to their tasks in
the target-application and to their progression.
Model of Articulation between aLDEAS Assistance Rules
17
Additionally, they must meet the pedagogical
objectives and strategies of teachers. In SEPIA, such
assistance systems require the definition of a lot of
rules which need to be articulated with different
modes. For instance, an assistance system provides
the learner with progressive assistance to solve an
exercise when he asks for assistance. For the first
time, this system gives an explanation, then a hint
and finally a solution. An explanation can be given
by showing the messages step by step. So the rules
defining this assistance system are articulated in
both progressive and successive modes. However,
aLDEAS and SEPIA have not yet provided an
explicit and easy way of definition of modes of
articulation between rules. To tackle this
problematic, we firstly made a state of the art in
order to identify existing modes of articulation
between assistance elements. We sum up this work
through the examples (section 3.1) and present the
support of these modes in some tools (section 3.2).
Then, in order to allow aLDEAS and SEPIA to
support these modes, we proposed a model of
articulation between rules (section 4) and its
implementation (section 5).We conclude this paper
by the presentation of the evaluations that we
defined for this research.
3 STATE OF THE ART
3.1 Modes of Articulation between
Assistance Elements
Currently, pedagogical assistance is found in some
applications. This assistance can be executed
according to different modes to sequence of
assistance events. These modes describe articulation
between assistance elements.
In many applications, an assistance element is
given independently from another. There are not
constraints between assistance elements. We can
easily find this mode in most applications with the
tooltips. A tooltip appears when the user hovers a
component on application interface. So, each tooltip
is independently showed. We take a concrete
example of IXL learning (IXL Learning, 2015) that
provides comprehensive, curriculum-aligned
mathematics and English content for preschool to
grade 12. It shows overviews of course through a
sequence of independent pop-ups (A in Figure 3)
when learners hovers links of course. We call this
mode of articulation independent mode.
The tutorials integrated in some applications
provide step by step assistance in order to guide
users. For instance, Connectify (B in Figure 3)
(Connectify, 2015) provides the messages one after
the others which allow user to learn the use of this
application. Such messages may also explain step by
step user errors. Each assistance element is
constraint by the end of the previous assistance
element. We call this mode of articulation successive
mode.
Hot Potatoes (Winke and MacGregor, 2001)
allows teachers to create different types of exercises.
It is especially useful for creating online, interactive
language learning exercises and for providing
pedagogical assistance such as diagnosis to verify
the answers of the learner. The diagnosis can be
given for a several parts at the same time. For
instance, an exercise created by EOLF in Franche-
Comté university (EOLF, 2016) (C in Figure 3),
shows correct or incorrect answers for an English
exercise at the same time. We can get the assistance
in the forms frequently offered on web where the
diagnosis on different user inputs can be
simultaneously displayed. These examples show that
all assistance elements can be simultaneously given.
We call this mode of articulation simultaneous
mode.
Figure 3: Examples of modes of articulation between
assistance elements.
DCCSEDU 2016 - Doctoral Consortium on Computer Supported Education
18
ActiveMaths (Melis et al., 2001) offers to
learners a mathematical web application with
pedagogical assistance. The learner can ask for
assistance in progressive way (D in Figure 3). At the
first request, a hint is given to the learner. Then, at
the next request, the given hint is more detailed. At
the last request, the solution is given in order to
avoid blocking the learner in working on problems.
The assistance is more and more detailed and
concrete. We call this mode of articulation
progressive mode.
Some applications consult information sources to
provide suitable assistance such as the application
state, the user profile or the user’s choice. Thus, both
ActiveMaths and Hot Potatoes allow consulting the
state of the application such as a text filled by the
learner. For instance, ActiveMaths (E in Figure 3)
verifies answers of the learner and shows the result
(syntax error, incorrect/correct answer).
Consultations are therefore essential to provide a
suitable assistance. We call this mode of articulation
interactive mode.
These modes of articulation can also coexist or
be combined in assistance systems. For instance,
tutorials provide step by step assistance, but in one
step, a textual message and an enhancing can be
simultaneously performed. This is a combination of
successive and simultaneous modes.
3.2 Related Works
The modes of articulation presented in the previous
section exist in many applications. However, some
tools allow also defining these modes explicitly or
implicitly. So, we confronted these modes to tools
related to our approach.
In Marco advisor systems (Richard and
Tchounikine, 2004), the advices are represented as a
graph. The subset of graph represents the assisted
website. There is no explicit articulation between
advices because each advice is represented
independently. However, the articulation between
the elements of a same advice can be conditioned by
the navigation history of the user. This is a case of
interactive articulation of assistance but implicitly
represented. In the Astus platform (Paquette et al.,
2014), educational interventions are represented as
rules organized in a graph. The conditions of the rule
make explicit interactive articulation between
interventions. In the same way, the Epitalk system
(Paquette et al., 1996) explicitly represents the
advices through a tasks graph. These works
concentrate on the definition of assistance but do not
explicitly address the aspect of articulation.
Therefore, the articulation may be defined explicitly
or implicitly by their tools and we can’t find the
presence of progressive and successive modes of
articulation.
In another way, the Grafcet graphical language
(David, 1995) is proposed in order to represent the
sequential automation in systems decomposable into
steps. Although Grafcet is not specific to assistance
systems, the expression of sequence of steps can
inspire our work. In Grafcet, a step can be an active
step, initial step, macro-step, etc. Actions are
associated with a step. The transition between two
steps is done through a transition. A transition is one
or more logical condition (boolean). With Grafcet,
we can describe explicitly the independent,
successive, interactive, simultaneous modes but only
implicitly the progressive mode. In addition, some
elements of Grafcet are not suitable with the ones of
aLDEAS. For instance, Grafcet doesn’t distinguish
events from conditions as aLDEAS does. It contains
also useless information in our context to the phase
of specification of an assistance system such as
active state on step.
To overcome these limitations in the literature,
we proposed a model of articulation between
aLDEAS rules and implemented in SEPIA. This
model and its implementation are presented in the
following.
4 MODEL OF ARTICULATION
BETWEEN ALDEAS RULES
If aLDEAS and its implementation in SEPIA already
allow the definition of the articulation between
assistance elements such as those presented in the
section 3 with aLDEAS rules, the expression of the
articulation between the rules is implicit and can be
complex to define for the assistance designer. To
more effectively operationalize these modes of
articulation in our propositions, it is necessary to
allow defining explicitly the articulation between
aLDEAS rules.
Thus, an assistance system is currently defined
in the AGATE project by a set of aLDEAS rules
always at the same level. In the aLDEAS rules
pattern (Fig. 2), the trigger event, the end event and
the trigger condition are central elements to form the
articulation between rules. For instance, we defined
two rules R
1
and R
2
which describe two successive
steps in the tutorial of Connectify. So, these rules are
articulated in successive mode. It means that R
2
is
launched at the end of R
1
. For this order of launch,
Model of Articulation between aLDEAS Assistance Rules
19
the trigger event of R
2
must be the event "end of R
1
".
On the one hand, we must carefully define elements
in the rules in order to ensure correct articulation
between them. On the other hand, we must examine
them in order to understand which mode of
articulation to choose. Therefore, this articulation
between rules is implicitly expressed and is
complexly defined with aLDEAS.
For these reasons, we propose to complete our
language by a model of articulation between
assistance rules. To simplify the representation of
the model, we note that rules between which we
want to make an articulation are named R
i
with i
[1, n], (n 2). The representation of our model is
given in Figure 4. It gives an overview of the five
modes of articulation that we identified from a study
of existing works: independent, successive,
simultaneous, progressive and interactive.
In each mode of articulation, there are constraints
that rules must respect to ensure the correct
articulation between them (for instance, for
successive mode, each rule should be launched by
the end of the previous rule). The constraints of each
mode of articulation are shown in the next section
with examples of assistance. These examples of
assistance are inspired by examples presented in
section 3. To simplify, we describe only three rules
articulated for each example.
4.1 Independent Mode of Articulation
In the independent mode of articulation (Figure 4)
the rules R
i
are launched by their own trigger events.
This mode doesn’t impose any constraint. The
definition of rules articulated in independent mode
reflects the classical definition with aLDEAS.
Obviously, the other modes presented thereafter are
specific cases of this mode with specific constraints
on rules.
Example A in Figure 5 is a case of assistance to
IXL Learning (section 3.1). Here, we present a
similar but simpler assistance written in aLDEAS
which takes only the three first overviews
corresponding to the three first courses. This
assistance is created with 3 rules articulated in
independent mode. The rules R
1
, R
2
, R
3
are
respectively corresponding to the three courses
“Counting review - 0 to 10”, “Count to fill a ten
frame”, “Counting review - up to 20”. Each rule
begins with its own trigger event “hover on link of
course in order to show a message which presents
Figure 5: Examples of model.
Figure 4: Model of articulation between aLDEAS
assistance rules.
DCCSEDU 2016 - Doctoral Consortium on Computer Supported Education
20
overview of this course.
4.2 Successive Mode of Articulation
In the successive mode of articulation (Figure 4), the
rules are launched one after the other, it means that
at the end of the rule R
i
, the rule R
i+1
is launched.
In the detailed definition of this mode of
articulation (Figure 6), we can see that the rule R
i
is
forced to have at least an end event and R
i + 1
is
forced to have a trigger event "end of R
i
." This
constraint is applied to all rules except the first and
last ones. The first rule R
1
can begin with any trigger
event(s) and the last rule R
n
may end with none, one
or several end events. In this mode of articulation,
the rule R
1
is an entry point of the rules R
i
. So, the
trigger events of the rule R
1
launch this set of rules
in successive mode.
However, a rule R
i + 1
cannot be launched until
the end of its preceding rule R
i
. Consequently, if R
i
has a trigger condition that is not validated at the
time of the assistance execution, the rule and its
assistance actions will not be executed until its end
events, and the following rule will therefore not be
launched. So, the whole sequence of rules is
interrupted. This requires a rule R
i
to contain a
condition to have an alternative "else". This
alternative ensures that the condition is always valid.
Figure 6: Constraints on rules of successive mode of
articulation in aLDEAS.
Let’s take the example of the assistance to
Connectify (section 3.1). Here, we present an
assistance similar but simpler which takes only the
three first steps of the tutorial of Connectify. This
assistance is created with three rules articulated in
successive mode. These three rules (defined in more
detail in Figure 7) must respect the constraints of the
successive mode (Figure 6). Thus, the first rule R
1
waits until a user’s click on bouton “Tutorial” in
order to show a message of welcome and closes this
message after 10 seconds. Then, the rule R
2
that
waits until the end of R
1
shows a message of internet
connection check and closes this message after 10
seconds. Finally, the rule R
3
that waits until the end
of R
2
shows a message of choice of an internet
connection as well as highlight the related combo
box. These message and highlight is also closed after
10 seconds.
Figure 7: Detail of three rules articulated in successive
mode in aLDEAS.
4.3 Simultaneous Mode of Articulation
The simultaneous mode of articulation (Figure 4)
allows executing several assistance rules
simultaneously.
In the simultaneous mode of articulation, the rule
R
i
must begin the same trigger events as ones
defined for this mode. When these events occur, all
rules are launched at the same time.
Let’s take the example of the assistance to an
exercise created with Hot Potatoes that shows
simultaneously all errors of answers. Here, we
present an assistance similar but simpler which takes
only the three first answers corresponding to three
first user inputs. This assistance is created with three
rules R
1
, R
2
and R
3
articulated in simultaneous mode
(Figure 5). These three rules must begin with trigger
event “click on bouton Check”. When the learner
clicks on this button, they are launched. They verify
learner’s answers with the consultation of the user
inputs on the application and shows the result
(correct or incorrect) by adding a text (OK if correct,
X else) near to these user inputs.
4.4 Progressive Mode of Articulation
In the progressive mode of articulation (Figure 4),
the launch of assistance rules depends on the number
Model of Articulation between aLDEAS Assistance Rules
21
of times that the learner is in a same situation. In
particular, this mode allows to provide the user with
the assistance more and more detailed and concrete
to meet a repeated request of assistance.
In the successive mode of articulation, the rule
R
1
is the entry point of assistance to successively
start the rules R
i
. In the simultaneous mode, all rules
R
i
start with same trigger events. However, in this
progressive mode, there must be an additional rule
as an entry point to constraint the launch of the rules
R
i
. We call this rule R’. R’ launches one rule among
the rules R
i
, according to the number of launches of
the rule R’. In the rule R’, each rule R
i
is associated
with an interval [left
i
, right
i
]. It means that R
i
is
launched for one or several times between left
i
and
right
i
. For the first times [left
1
, right
1
], R’ launches
R
1
and for the next times [left
2
, right
2
] R' launches
R
2
. In this mode of articulation, the rules R
i
must
begin with a trigger event "launch by a rule (R’)".
Let’s take the example of the assistance to
ActiveMaths (section 3.1) that gives at first a hint,
then a more detailed hint and finally the solution
when the user repeatedly asks for assistance. The
assistance is created with three rules articulated in
progressive mode (Figure 5). R’ begins with trigger
event “click on bouton Help” and launches the rules
R
1
, R
2
, R
3
which show respectively a hint, another
more detailed hint and the solution. This launch is
constrained by the number of launches of R’. It
means that the number of clicks on button “Help” is
counted. To be launched, these three rules R
1
, R
2
and
R
3
must begin with a trigger event “launch by a rule
(R’)”. So, R
1
is launched by R’ for the first click on
bouton “Help”, R
2
for the second click and R
3
for
the third click.
4.5 Interactive Mode of Articulation
In the interactive mode of articulation (Figure 4),
one of the rules R
i
is launched according to a
consultation of the user profile, of the application
state, of the history of the assistance, of the trace and
/ or of the user.
Again, a rule R’ is used as an entry point for the
launch of the assistance. Each rule R
i
is associated
with an alternative of the trigger condition of R’. R
i
must begin with the trigger event "launch by a rule
(R’)." The progressive mode of articulation (see
section 5.3) is a special case of the interactive mode,
frequently encountered in the existing assistance
systems and in which the trigger condition of R’ is
exclusively a number of launches of R’.
Let’s take the example of the assistance to
ActiveMath (section 3.1) which shows the diagnosis
by consulting the learner’s user. This assistance is
created with three rules articulated in the interactive
mode. The representation of the additional rule R’ is
the same as the representation of interactive
articulation. R begins with the trigger event “click
on bouton Check” and launches the rules R
1
, R
2
and
R
3
which show respectively a syntax error, a
calculation error and success. This launch is
constrained by the learner’s answer: the value of the
text box entered by the learner. To be launched,
these three rules R
1
, R
2
and R
3
must begin with a
trigger event “launch by a rule (R’)”. When the
learner clicks on bouton “Check”, one rule among
the three rules is launched by R’. If the entered value
of the text box does not belong to float type, R
1
is
launched, if this value is equal to 1, R
3
is launched
and elsewhere, R
2
is launched.
5 IMPLEMENTATION OF OUR
MODEL OF ARTICULATION
BETWEEN RULES
We implemented this model of articulation between
rules in SEPIA that haven’t supported the explicit
expression of the articulation until now (Figure 8).
More concretely, we enriched the SEPIA assistance
editor to support designers to define explicitly the
five modes of articulation as well as to facilitate
their definition. In order to facilitate the
comprehension of designers, we adopt the notion of
bloc that regroups the rules articulated in a given
mode among these five modes.
Figure 8: SEPIA completed with the model of articulation
between rules.
Thus, SEPIA allows assistance designers to
define and view graphically blocs of rules articulated
in the wished mode. Each mode has constraints on
rules to ensure a correct articulation. The automatic
application of these constraints facilitates the
definition of assistance systems. It allows
automatically generating or modifying aLDEAS
rules as well as eventually verifying designer’s
definition. For instance, to define a bloc of rules
articulated in the successive mode, the next rules
must have a trigger event “end of previous rule”.
DCCSEDU 2016 - Doctoral Consortium on Computer Supported Education
22
The definition of the bloc will add automatically this
event in these rules. Otherwise, the previous rules
must have at least an end event which must be
determined by the designers. Therefore, SEPIA will
check this constraint and show a message if it is not
satisfied.
An assistance system is now represented in
SEPIA by rules not only defined by designers but
also generated and modified by the application of
constraints thanks to blocs (Figure 8). This allows
keeping the current operation of the SEPIA engine
which doesn’t need to change because of the
implementation of model of articulation. This
implementation allowed us to create and executes
with SEPIA examples of assistance similar to
assistance presented in the section 3. Figure 9 shows
the execution of these examples.
Figure 9: Execution of examples of assistance created by
SEPIA with the implementation of our model.
6 EVALUATION
The aLDEAS language and its implementation in the
SEPIA system were previously evaluated. The
usability of aLDEAS and the usability of the SEPIA
were evaluated in (Ginon et al., 2014b). The
execution of assistance systems with the SEPIA in
ILEs was evaluated in (Ginon et al., 2014c).
In this paper, we propose a model of articulation
between assistance rules. Therefore, in this section,
we focus on the evaluation of this model of
articulation. Regarding the feasibility of model, it is
demonstrated by the implementation of our model in
SEPIA. The examples in Figure 9 show the
possibility of our model. Thus, the model of
articulation allows us to define the assistance
systems similar to ones presented in section 3. In
addition, we made an experiment of our model with
few users and will make another with more users in
spring 2016. The objective of these experiments is to
evaluate: (C1) the capacity of comprehension of an
assistance system created by using blocs, (C2) the
capacity of use of blocs, (C3) the benefices of use of
blocs, (C4) the coverage of 5 modes relative to
expectation of designers. The designers must work
with three ways of definition of assistance systems:
definition without bloc by using the textual
interface, definition without bloc by using graphical
interface and definition with bloc by using the
graphical interface. With each way, they will
execute 3 steps: comprehension, modification, and
completion of an assistance system. Next, they must
create an assistance system with a preferred way of
definition. Each user uses one mode of articulation.
We started the first experiment with students in
France in order to observe and improve the future
experiments. In this experiment, there are 5 master
students. We only observed their tasks without
considering the result. However, the majority of
them answered that they understood how an
assistance system operates and define an assistance
system with bloc. Then, we improved the documents
for the experiment with 6 Vietnamese students in the
course HCI (Human Computer Interaction) in
Vietnam. We summarized results, which is
presented through the above figures. Figure 10
shows the number of users who succeeded the
comprehension, modification and completion by
three ways (for evaluation of C1, C2). There are no
major difference between them. However, most
users (5 out of 6 users) prefer to use the bloc in order
to define an asked assistance system (Figure 11). In
more detail, these users indicate that the
comprehension and the definition of an assistance
system with bloc are easier than others (Figure 12)
(for evaluation of C3). The five modes are indicated
enough for the definition of an assistance system
because the users didn’t give any other mode
existing in other applications or in reality (for
evaluation of C4). Through this experiment, we can
think that our model of articulation facilitates the
comprehension, the definition of an assistance
system. The modes of articulation deducted from
bibliographical studies (cf. section 3) can define
Model of Articulation between aLDEAS Assistance Rules
23
various assistance systems.
However, the number of users who participated
in the above experiment is low. Therefore, we will
make another experiment in spring 2016 with 30
French master students with the same objectives. For
the evaluation of coverage of model, we will
improve this experiment by asking students to
imagine a pedagogical assistance system. They must
show whether it can be defined by using blocs with
one among the five modes of articulation or with a
non-existing mode of articulation.
Figure 10: Number of success for the realization of tasks
in our experiment.
Figure 11: Levels of comprehension and definition of an
assistance with three ways of definition.
Figure 12: Number of users for their preferred way of
definition out of 6 tests.
7 CONCLUSION AND FUTURE
WORK
In this article, we presented the model of articulation
between aLDEAS assistance rules which completes
the aLDEAS language. This model explicitly
expresses the notion of articulation between rules of
an assistance system. It offers five modes of
articulation corresponding to those we have
identified in our bibliographical study: independent,
successive, simultaneous, progressive, and
interactive. We implemented this model in the
SEPIA system by adding the notion of bloc of rules
articulated in a mode. This implementation have two
main advantages: it makes explicit the definition of
blocs of rules with graphical interface and it applies
semi-automatically constraints on rules. With the
introduction of this model in our approach, an
assistance system is defined not only by a set of
rules, but also by a set of blocs that explain the
articulation between these rules. It allows teachers to
view more explicitly as well as define more easily a
complex assistance system. We evaluated our
propositions by the experiment which showed some
big potentials.
However, an assistance system can be described
by many blocs of rules articulated in different
modes. SEPIA just shows the graphical
representation of a bloc but not the global graphical
representation of all the blocs. The blocs are listed in
a table that limits designer’s view of a whole
assistance system. Therefore, in the future, we will
aim at a global graphical representation of assistance
systems which will be more intuitive.
As part of thesis, we continue to evolve SEPIA
which will facilitate the definition of pedagogical
guidance. We will find out how existing applications
or systems propose pedagogical activities suitable to
learners. For example, the activities can be
temporally planned or the proposition of activities
can be constraint by states of previous activities (e.g.
remediation activities). Then, with SEPIA, we try to
define assistance systems which can also propose
these activities in order to identify difficulties. Thus,
SEPIA has not yet supported the concepts
“pedagogical guidance” and “learning activity”. It’s
difficult and complex for assistance designers who
wish to define a pedagogical guidance. So, we aim
to propose these concepts in SEPIA. Through our
state of the art, we will enrich these concepts in
SEPIA (for example, temporal attribute in
pedagogical guidance, output states in pedagogical
activity).
REFERENCES
Antoniadis, G., Echinard, S., Kraif, O., Lebarbé, T.,
Loiseau, M., & Ponton, C., 2004. NLP-based scripting
for CALL activities. In Proceedings of the Workshop
on eLearning for Computational Linguistics and
DCCSEDU 2016 - Doctoral Consortium on Computer Supported Education
24
Computational Linguistics for eLearning pp. 18–25.
AGATE. http://liris.cnrs.fr/agate/ [retrieved 2015]
Connectify. http://www.connectify.me/ [retrieved 10
December 2015]
David, R., 1995. Grafcet: A powerful tool for specification
of logic controllers. In Control Systems Technology,
IEEE Transactions on, vol. 3, pp. 253–268.
EOLF. http://eolf.univ-fcomte.fr/wp-content/uploads/gram
mar/verbs/irregular/10.htm [retrieved 10 December
2015]
Gapenne, O., Lenay, C., Boulier, D., 2002. Defining
categories of the human/technology coupling:
theroretical and methodological issues. In workshop
ERCIM on User Interface for All, pp. 187-198. France.
Ginon, B., Jean-Daubias, S., Champin, P.-A., & Lefevre,
M., 2014a. Setup of epiphytic assistance systems with
SEPIA. In: EKAW, pp. 1-4. Linkoping.
Ginon, B., Jean-Daubias, S., Champin, P.-A., & Lefevre,
M., 2014b. aLDEAS: a Language to Define Epiphytic
Assistance Systems. In: EKAW, pp. 153-164.
Linkoping.
Ginon, B., Thai, L.-V., Jean-Daubias, S., Lefèvre, M., &
Pierre-Antoine, C., 2014c. Adding epiphytic assistance
systems in learning applications using the SEPIA
system. In EC-TEL, pp. 138-151. Graz.
IXL learning, https://eu.ixl.com/math/grade-1 [retrieved
10 December 2015]
Melis, E., Andres, E., Budenbender, J., Frischauf, A.,
Goduadze, G., Libbrecht, P., Pollet, M. & Ullrich, C.,
2001. ActiveMath: A Generic and Adaptive Web-
Based Learning Environment. In: IJAIED, vol. 12, pp.
385-407.
Paquette, G., Pachet, F., Giroux, S., Girard, J., 1996.
Epitalk, a generic tool for the development of advisor
systems. In IJAIED, p. 349-370.
Paquette, L., Lebeau, J.-F., Beaulieu, G., & Mayers, A.,
2014. Designing a Knowledge Representation
Approach for the Generation of Pedagogical
Interventions by MTTs. In International Journal of
Artificial Intelligence in Education, p. 1-39.
Richard, B., & Tchounikine, P., 2004. Enhancing the
adaptivity of an existing website with an epiphyte
recommender system. In New review of hypermedia
and multimedia, vol. 10, p. 31-52.
Winke, P., & MacGregor, D., 2001. Hot Potatoes version
5. In: Language Learning Journal, vol. 24, pp. 30–33.
Model of Articulation between aLDEAS Assistance Rules
25