PERSPECTIVES AND ASPECTS IN POEML

Supporting Adaptation in Learning Processes

Manuel Caeiro, Luis Anido and Martin Llamas

Department of Telematic Engineering, University of Vigo, Vigo, Spain

Keywords:

Model-driven engineering, Process-based Modeling, e-Learning, Educational modeling languages.

Abstract:

Adaptation is a key feature of e-learning systems. The educational process performed by the support of e-

learning systems should be carried out taking into account the cognitive and other particular characteristics

of the students, of the teachers and of the environments. However, this adaptation is not a simple process.

It implies the conjunction of technical and pedagogical issues. This paper introduces a solution to support

adaptation in the context of an Educational Modeling Languages (EML). EMLs have been conceived to allow

us the modeling of learning units, enabling the expression of different pedagogical approaches. Eventually,

these models are intended to control and support the development of the corresponding learning units.

1 INTRODUCTION

During the last years there have been several initia-

tives trying to support e-learning based on a model-

oriented approach. Instructional Engineering has

been proposed by (Paquette, 2004) as ”a method that

supports the analysis, the creation, the production,

and the delivery of a learning system, integrating

the concepts, the processes, and the principles of in-

structional design, software engineering, and knowl-

edge engineering”. A main result of this initiative

is a proposal for a modeling language in accordance

with the previous description. Similarly, Educational

Modeling Languages (EMLs) have been proposed re-

cently as process-based modeling languages to de-

scribe learning units and support their execution (Vi-

gnollet et al., 2006). More speciﬁcally, IMS Learning

Design (IMS-LD) is the current de-facto EML stan-

dard (Koper et al., 2003). This language has been

developed as a computational EML that can be used

by suitable Learning Management Systems (LMSs) to

support, control and manage the development of e-

learning experiences (Koper and Olivier, 2004). A

review of these and other modeling languages from a

graphical point of view has been recently performed

in (Botturi and Stubbs, 2007).

Despite the existence of numerous initiatives and

proposals, the support of e-learning based on the

model-oriented approach has not succeeded, yet. A

main problem in this approach is the difﬁculty to

create models of learning units by ﬁnal users. The

achievement of a single language that enables to cre-

ate models in accordance with different pedagogical

approaches is feasible (at least following a process-

based approach, as IMS-LD), but it is not very usable.

This paper introduces a new EML to sim-

plify the modeling of learning units named as

Perspective-oriented EML (PoEML). Following a ba-

sic separation-of-concerns idea, PoEML approaches

the modeling of learning units through several per-

spectives and aspects. Each perspective and aspect

involves the modeling of a part of the whole learning

unit in a way (almost) separated from the other per-

spectives. As a result, these separation into perspec-

tives brings many beneﬁcts. Particularly, this paper is

focused on the description of the support of adapta-

tion based on perspectives and aspects.

2 POEML

This section introduces a new EML devoted to in-

crease the expressiveness of existing EMLs while ob-

serving important capabilities such as adaptability. It

has been adopted a separation of concerns approach.

Basically, the modeling of educational units is not

attained as a single and whole stuff, as in existing

EMLs, but the learning units are observed from sev-

eral perspectives that are modeled separately . Fol-

lowing this approach the new EML has been named

as Perspective-oriented EML (PoEML). Next sections

226

Caeiro M., Anido L. and Llamas M. (2009).

PERSPECTIVES AND ASPECTS IN POEML - Supporting Adaptation in Learning Processes.

In Proceedings of the 4th International Conference on Software and Data Technologies, pages 226-231

DOI: 10.5220/0002258202260231

 SciTePress

describe the PoEML language at abstract and struc-

tural levels.

2.1 Foundations

The key strategy for the development of Po-

EML has been the separation-of-concerns principle.

Separation-of-concerns is a long-standing idea that

simply means that a large problem is easier to manage

and solve if it can be broken down into parts and each

part can be approached separately. It is an important

design approach in many areas, such as software de-

sign (Parnas, 1979), used to facilitate the understand-

ing, design and management of complex systems. In

addition, UML is a modeling language for software

engineering where different diagrams are proposed to

model different issues: use cases, analysis, design,

etc. A similar example in other domain is the architec-

tural plans, which follow a separation of concerns for

the development of buildings. In the learning domain

there are also some proposals in which a certain kind

or separation of concerns is provided, such as (Strij-

bos et al., 2004), where learning units are considered

through several orthogonal axis. Anyway, as long as

we know, PoEML is the ﬁrst attempt that takes the

separation of concerns as driven development princi-

ple.

Another important foundation of PoEML is Activ-

ity Theory (Engestrom et al., 1999). This theory has

been used to analyze the issues involved in learning

units and to identify an appropriate separation of con-

cerns (Caeiro-Rodr´ıguez et al., 2007). The Activity

Theory has guided us in the separation-of-concerns

in learning units towards the identiﬁcation of 13 per-

spectives and 4 aspects. They have already been de-

scribed in a previous document (Caeiro-Rodr´ıguez,

2007).

2.2 The PoEML Conceptual Model

The PoEML conceptual model describes the main

concepts of the language and the relations among

them. From a conceptual point of view PoEML is

mainly characterized by two important issues: the

adoption of a special kind of task concept as basic

building block and its hierarchical and structured na-

ture. In addition, there is one root element with spe-

cial components. Next sections describe these impor-

tant features.

2.2.1 The Basic Concept

PoEML is a process-based language, as other EMLs.

Nevertheless, the basic concept of PoEML is not the

task, but the Educational Scenario (ES). There are

two main differences that distinguish the PoEML ES

and the typical EML task concepts:

• The Goal is recognized as a ﬁrst-class entity in Po-

EML. It includes its own identiﬁer, sub-elements

and control information, as if it is compulsory

or optional, input/output parameters, etc. By the

contrary, in existing EMLs goals are represented

just as a textual description that informs partici-

pants about what they have to do at each task. In

this way, there is no control information associ-

ated with the goal. Actually, this control informa-

tion is available in the models of learning units,

but in the relations among tasks. More important,

this informationis completely interleavedwith the

speciﬁcation of the learning unit structure. In this

way, it is very difﬁcult and complex to change

the goals of a learning unit without changing the

structure.

• Second, the ES includes not just elements

(namely: Goals, Roles, Environments, Data El-

ements), but also speciﬁcations related with the

control, management and coordination of such el-

ements. Particularly, an ES includes speciﬁca-

tions to support the modeling of the issues such

as the assignment of permissions to participants,

the notiﬁcation of events, the invocation of oper-

ations, the ordering among Sub-ESs and the tem-

poral plan of Sub-ESs. The control involved in

these speciﬁcations affects only to the elements

included in the ES.

In this way, the ES provides a completely encap-

sulated model that facilitates a separation of concerns

approach to the modeling of learning units. On the

one hand, an ES does not make reference to any issue

outside of its own contained elements. On the other

hand, each one of these elements and speciﬁcations is

considered at a different perspective. In practice, the

model of an ES is made up by the sub-models of its

Goals, Roles, Environments, Data Elements and spec-

iﬁcations of Authorization, Awareness, Interaction,

Order and Temporal. In addition all these elements

and speciﬁcations are modeled as ﬁrst-class entities,

each one of them involving its own unique identiﬁer,

sub-elements and control information. This model

is completetly different from the solution adopted on

IMS-LD, where the several concerns are interleaved.

Figure 1 illustrates this ES concept. It is shown

how an ES can include several elements: Goals (at

least 1), Roles, Environments, Data Elements and

Causal Descriptions; and speciﬁcations: Awareness,

Authorization, Interaction, Order and Temporal. It is

important to stress that the modeling of each one of

these elements and speciﬁcations can be performed

PERSPECTIVES AND ASPECTS IN POEML - Supporting Adaptation in Learning Processes

227

using different PoEML packages, namely: Func-

tional, Participants, Environments, Data, Authoriza-

tion, Awareness, Interaction, Order and Temporal. In

the ﬁgure, it is also shown that Goals, Roles and En-

vironments can also include Data Elements.

Educational Scenario

Goal

Role

Goal

Data

Element

Specification Specification Specification Specification

Causal

Role

Environment Goal

Data

Element

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

Order

Specification

Temporal

Specification

1..*

*

* * * * *

Elements

Specifications

*

Figure 1: The ES elements and structure.

2.2.2 The Hierarchical and Structured Modeling

of Learning Units

This conceptual view of PoEML is completed by con-

sidering the hierarchical aggregation of ESs to model

learning units. Basically, any model of learning unit

is made up by several ESs that are aggregated in a

hierarchical way. As it is stated in the previous sec-

tion, each ES includes its own Goals, Roles, Envi-

ronments, Data Elements and Speciﬁcations. The hi-

erarchical aggregation of ESs indicates that a certain

ES (parent-ES) can be made up by several ESs (sub-

ESs or children ESs). This hierarchical structure of

the learning unit can be speciﬁed using the PoEML

Structural package.

In addition to the hierarchical structure, a learn-

ing unit can require the establishment of relations be-

tween Goals, Roles, Data Elements, sub-ESs and En-

vironments. These relationships have been identiﬁed

in the separation-of-concerns analysis as particular

ﬂows:

• The Functional Flow involves the relations that

can be established among Goals included in dif-

ferent ESs. These relations can be of two different

types:

– Completion relations. They can be used to in-

dicate what conditions over the state of some

Goals have to be satisﬁed to complete a cer-

tain Goal (generally this Goal is contained in

a parent-ES and the other Goals in its sub-ESs).

For example, to obtain the car driven license is

required to pass the theoretical test and to have

succeeded on the practical exam. Another ex-

ample, to pass the theoretical test it is required

to pass at least three of the four parts in which

the test is divided.

– Attempt relations. These relations can be used

to indicate what conditions over the state of

some Goals have to be satisﬁed to enable the

attempt of other Goal (generally all these Goals

are contained in sub-ESs of the same parent-

ES). For example, the practical exam can be

attempted when the theoretical test has been

passed.

• The Participants Flow involves the relations that

can be established among Roles included in dif-

ferent ESs. They are used to represent the transfer

of participants among Roles. For example, the as-

signment of learners to a Role ”team”, mada up

by 4 learners.

• The Data Flow involves the relations that can be

established among Data Elements, indicating the

transfer of data among them. For example: the

grade obtained in the assessment of a question-

naire has to be transferred to the next ES and also

maintained in the transcript of the learner.

• The Control Flow involves the relations that can

be established among the sub-ESs of a certain

ES. Basically, this ﬂow indicates what sub-ES has

to/can be done at a certain moment. It can be de-

termined in two different ways:

– In relation with the state of performance of

other ESs. For example, when the learner pass

the theoretical exam she can initiate any of the

practices assigned. Then, when the learner ﬁn-

ishes all the practices, she has to perform a

practical exam.

– In accordance with a temporal planiﬁcation.

For example, the presentation of the subject is

going to be produced on March 11 at 5pm. An-

other example, the ﬁnal exam has to be per-

formed in less than 1 hour.

• There is not any Environment Flow, as the con-

tents of the Environments cannot be transferred,

but it is possible to establish relations between En-

vironments. For example, the lab used to perform

the ﬁnal practice exam has to be the same lab in

which the learners performed all the practices.

We would like to stress that the modeling of each

one of these issues can be performed using a different

PoEML package, producing as a result a speciﬁc sub-

model. In addition, this modeling is done in a very

structured way, as the relations have to be established

among elements included in the same ES or in its sub-

ESs.

This hierarchical and structured approach is rep-

resented in Figure 2. There is as a main block, repre-

senting the Root ES, that contains two blocks repre-

senting sub-ESs: ES A and ES B. In addition, ES A

ICSOFT 2009 - 4th International Conference on Software and Data Technologies

228

contains three further sub-ESs: ES A.1, ES A.2 and

ES A.3. The ﬁgure shows this structural arrangement

and the relations established between Goals, Roles

and Data Elements. It is quite obvious that the view

in a single picture of all these issues is very confus-

ing and complex. The PoEML separation on several

perspectives provides a clearer view, enabling us to

observe the different issues and relations in separated

sub-models.

2.2.3 The Root ES

To complete this conceptual model it is necessary to

introduce the concept of Root ES. The Root ES rep-

resents the global learning unit. It can include the

same elements than any other ES. In addition, as it

represents the global learning unit, it can also include

further elements that capture the connections of the

learning unit with the ”outside world”. This outside

world is composed by real resources, tools, partici-

pants, environments and organizations that exist out-

side of the learning unit. For example, a learning unit

model should enable to indicate that a certain PDF

document has to be used. The connections between

these external elements and the model of learning unit

are established in the Root ES and they can be refer-

enced from any sub-ES of the Root ES, independently

of its depth level in the hierarchical ESs’ structure.

The speciﬁcation of tools and organizations involves

special features and it is performed in speciﬁc per-

spectives.

In this way, from a conceptualpoint of view, an ES

is an encapsulated entity including all the elements

that made up it and the speciﬁcations that deﬁne its

structure and behavior. Furthermore, a root-ES is a hi-

erarchical and structured model that can represent any

piece of learning at different levels of aggregation,

from simple lessons to complete curricula, and that

includes all the connections with the ”outside world”

required to perform its ”enactment”.

3 ADAPTATION

Adaptation is a central subject in e-learning and it can

be considered in many different ways. In general, e-

learning adaptation is focused on the optimization of

certain measures (e.g., learning time, economic costs,

user satisfaction) in order to improve the effectiveness

and efﬁciencyof the educationalprocess. This section

consider adaptation in the general e-learning domain

ﬁrstly and then in PoEML.

3.1 Adaptation in e-Learning

Traditionally, adaptive hypermedia has captured the

efforts in the e-learning adaptation domain (Henze

and Nejdl, 2004), focusing on the student and on

the contents (Brusilovsky, 2000) (Weber and Specht,

1997) (Bra et al., 2003). These proposals are based

on the model of a single learner that is provided with

contents in accordance to her/his characteristics and

progress (determined through the grades obtained in

tests and questionnaires). Nevertheless, this is just

one of the approaches that can be followed to achieve

learning. As in conventional face-to-face educational

settings, many current e-learning systems are build

on the basis of different pedagogical aproaches, such

as constructivist or social ones. These systems pro-

vide functionalities that allow learners to communi-

cate with other learners and teachers, to play, to exper-

iment, to perform authentic activities, etc. Therefore,

in this kind of e-learning systems, a broader view of

adaptation should be supported, taking into account

not just the contents and learner characteristics, but

all the elements involved: participants, communica-

tions, activities, context, etc.

The search of a general solution to support adap-

tation in e-learning needs to take into account the va-

riety of issues and approaches already considered. An

interesting classiﬁcation framework to analyze adap-

tation in e-learing is described in (Specht and Burgos,

2007) based on four main questionsis:

• What parts or components in the learning process

are adapted? This question focuses on the part of

an adaptive application that is adapted. Examples

can be to personalize the pace of the instruction,

adaptation of content presentations, the sequenc-

ing of contents, etc.

• What information does the system use for adap-

tation? In most proposals a learner model is the

basis for the adaptation, taking into accout her/his

knowledge, preferences, interests or cognitive ca-

pabilities. Nevertheless there are examples where

the adaptation takes place to the learner setting,

accessing devices, etc. Especially in social soft-

ware the information for adaptation can come out

of external information resources, collective log-

ging information, or even contextual sensor infor-

mation like the location of a learner.

• How does the system gather the information to

adapt to? There are a variety of methods to collect

information about learners to adapt to. Mainly im-

plicit and explicit methods like described in works

from user modeling can be distinguished.

• Why does the system adapt? This question mainly

PERSPECTIVES AND ASPECTS IN POEML - Supporting Adaptation in Learning Processes

229

Role

Environment

Goal

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

Order

Specification

Temporal

Specification

Role

Environment

Goal

Data

Element

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

Order

Specification

Temporal

Specification

Role

Environment

Goal

Data

Element

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

Data

Element

Root ES

ES A

ES B

Legend

Control Flow

Data Flow

Functional Flow

Participant Flow

Role

Environment Goal

Data

Element

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

ES B

Role

Environment Goal

Data

Element

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

ES B

Role

Environment Goal

Data

Element

Causal

Description

Authorization

Specification

Awareness

Specification

Interaction

Specification

ES B

ES A.1 ES A.2 ES A.3

Tool

Organization

Figure 2: The modeling of learning units using an hierarchical aggregation of ESs.

focuses on the pedagogical models behind the

adaptation. Classical educational hypermedia sys-

tems mainly adapt according for compensation of

knowledge deﬁcits, ergonomic reasons, or adapt

to learning styles for an easier introduction into a

topic.

3.2 Supporting Adaptation in PoEML

The support of adaptation in EMLs has been a main

topic. IMS-LD is able to manage different types of

adaptation (Burgos et al., 2006): Learning ﬂow, con-

tent adaptation, evaluation and interactive problem

solving support are best supported. Group adaptation

is supported via administrative tools for user group-

ing and group properties and modiﬁcation of a course

on-the-ﬂy can be partially implemented based on cur-

rent runtime environment restrictions. Also interface-

based adaptation is possible as long as the modiﬁca-

tions are made inside the Unit of Learning and not in

the player tool itself. Nevertheless, the provided so-

lutions are complex because the models of learning

units in IMS-LD have mane interdependences.

The support of adaptation in PoEML has been ar-

ranged in accordance with the previous four ques-

tions. In this way, the adaptations can be modeled

in clear and simple ways. Basically, there is a direct

relatioship among the language structure and the con-

sidered questions that facilitate the speciﬁcation:

• What parts or components in the learning process

are adapted? In PoEML these parts and compo-

nents are given by the several perspectives (except

the causal perspective). Each one of the perspec-

tives involves the structures and control that can

be modiﬁed.

• What information does the system use for adap-

tation? The information used for adaptation is

included in the four aspects. Depending of the

aspect it can be used data elements, events or de-

cisions.

• How does the system gather the information to

adapt to? In the case of data the gathering of in-

formation is performed using the data perspective.

In the case of events, it has to be used the aware-

ness perspective. In case of decisions, the way in

which information is collected is managed by ex-

ternal tools (e.g., a voting application) that can be

managed by the tools perspective.

• Why does the system adapt? The rationale about

the adaptation can be maintained in the causal per-

spective.

4 CONCLUSIONS

The main approach adopted in PoEML is the

separation-of-concerns. As in other engineering dis-

ciplines, the management of complexity is a key

point to facilitate the solution of problems and the

separation-of-concerns is a well-known technique to

tackle with it. This strategy enables us to focus the

attention on one perspective at each time, while ab-

stracting from others. In this way, PoEML is not pro-

ICSOFT 2009 - 4th International Conference on Software and Data Technologies

230

vided as a single set of elements, constraints and rela-

tionships, but it is made up by several sub-languages

(packages), each one of them involving particular ele-

ments, constraints and relationships focused on a cer-

tain perspective.

The paper devotes a special attention to the sup-

port of the variety of adaptation. The separation-of-

concerns approach adopted in PoEML is particularly

appropriate to translate the general ideas managed

about adaptation in e-learning to EMLs. In relation

with other EMLs, uniquely IMS-LD has considered

the support of adaptation. At this point, the solution

adopted in PoEML is much more simple and power-

ful.

ACKNOWLEDGEMENTS

This work has been funded by the Spanish Ministe-

rio de Educacion y Ciencia under grant TIN2007-

68125-C02-02,and by the Galician Conselleria de In-

novacion e Industria under grant PGIDIT06PXIB32

2270PR. It is also supported by the eContentplus

programme ECP 2007 EDU 417008 (www.aspect-

project.org), a multiannual Community programme to

make digital content in Europe more accessible, us-

able and exploitable.

REFERENCES

Botturi, L. and Stubbs, T., editors (2007). Handbook of

Visual Languages for Instructional Design: Theories

and Practices. Idea Group Inc. (IGI), Hershey, PA,

USA.

Bra, P. D., Aerts, A., Berden, B., de Lange, B., Rousseau,

B., Santic, T., Smits, D., and Stash, N. (2003). Aha!

the adaptive hypermedia architecture. In HYPER-

TEXT ’03: Proceedings of the fourteenth ACM con-

ference on Hypertext and hypermedia, pages 81–84,

New York, NY, USA. ACM.

Brusilovsky, P. (2000). Adaptive hypermedia: From intelli-

gent tutoring systems to web-based education. In ITS

’00: Proceedings of the 5th International Conference

on Intelligent Tutoring Systems, pages 1–7, London,

UK. Springer-Verlag.

Burgos, D., Tattersall, C., and Koper, R. (2006). Repre-

senting adaptive elearning strategies in ims learning

design. In International Workshop in Learning Net-

works for Lifelong Competence Development, TEN-

Competence Conference, Soﬁa, Bulgary.

Caeiro-Rodr´ıguez, M. (2007). Handbook of Visual Lan-

guages for Instructional Design: Theories and Prac-

tices, chapter PoEML: A Separation-of-Concerns Pro-

posal to Instructional Design, pages 185–209. Infor-

mation Science Reference.

Caeiro-Rodr´ıguez, M., Marcelino, M. J., Llamas-Nistal, M.,

Anido-Rif´on, L. E., and Mendes, A. J. (2007). Sup-

porting the modeling of ﬂexible educational units po-

eml: A separation of concerns approach. Journal of

Universal ComputerScience, 13(7):980–990.

Engestrom, Y., Miettinen, R., and Punamaki, R.-L., editors

(1999). Perspectives in activity theory. Cambridge

University Press, New York, USA.

Henze, N. and Nejdl, W. (2004). A logical characteriza-

tion of adaptive educational hypermedia. Hyperme-

dia, 10(1):77–113.

Koper, R. and Olivier, B. (2004). Representing the learning

design of units of learning. Educational Technology

and Society, 7(3):97–111.

Koper, R., Olivier, B., and Anderson, T., editors (2003).

IMS Learing Design Information Model. IMS Global

Consortium, New York, NY, USA.

Paquette, G., editor (2004). Instructional Engineering in

Networked Environments. Pfeiffer, San Francisco,

CA, USA.

Parnas, D. L. (1979). On the criteria to be used in decom-

posing systems into modules, pages 139–150. Your-

don Press, Upper Saddle River, NJ, USA.

Specht, M. and Burgos, D. (2007). Modeling adaptive edu-

cational methods with ims learning design. Journal of

Interactive Media in Education (Adaptation and IMS

Learning Design Special Issue), (8).

Strijbos, J. W., Martens, R. L., and Jochems, W. M. G.

(2004). Designing for interaction: six steps to design-

ing computer-supported group-based learning. Com-

put. Educ., 42(4):403–424.

Vignollet, L., David, J.-P., Ferraris, C., Martel, C., and Leje-

une, A. (2006). Comparing educational modeling lan-

guages on a case study. In ICALT ’06: Proceedings of

the Sixth IEEE International Conference on Advanced

Learning Technologies, pages 1149–1151, Washing-

ton, DC, USA. IEEE Computer Society.

Weber, G. and Specht, M. (1997). User modeling and adap-

tive navigation support in www-based tutoring sys-

tems. In Proceedings of the 6th international confer-

ence on user modeling, pages 289–300.

PERSPECTIVES AND ASPECTS IN POEML - Supporting Adaptation in Learning Processes

231