m-AdaptWeb

: AN ADAPTIVE E-LEARNING ENVIRONMENT

FACING MOBILITY

Adaptation and Recomendation Processes based on Context

Isabela Gasparini

1,2

, Ana Marilza Pernas

1,3

, Marcelo S. Pimenta

, José Palazzo M. de Oliveira

Avanilde Kemczinski

and Gerson Geraldo H. Cavalheiro

Instituto de Informática, Universidade Federal do Rio Grande do Sul, Porto Alegre, Brazil

Universidade do Estado de Santa Catarina, Santa Catarina, Brazil

Universidade Federal de Pelotas, Rio Grande do Sul, Brazil

Keywords: Web-based Learning, Mobile Learning, Adaptation, Situation-aware Recommendation.

Abstract: This work presents a case-study focused in practical learning scenarios of m-AdaptWeb, an adaptive

e-learning environment envisioning the exploration of mobile functionalities in the web-based learning

environment AdaptWeb

. The main objective of our research is to provide content adapted to different

students facing different learning problems when using their mobile devices. In this paper we present an

infrastructure to allow mobile-aware behaviour to an e-learning system and a case-study that emphasizes the

enrichment needed to be applied to learning objects metadata, in order to be effective in specific mobile

conditions. Also, we show how m-AdaptWeb provides support for the adaptations presented to the students

and for the recommendations depending on their progress in a specific learning activity.

1 INTRODUCTION

The growing adoption of distance learning (or e-

learning) brought new focuses and resulted in

fundamental changes in teaching and learning. In

particular, as e-learning environments become

widely used, more sophisticated personalized

features have to be investigated.

An efficient solution for distance learning are the

Adaptive Hypermedia Systems (AHS), which build

a model of goals, preferences and knowledge of each

individual user, and use this model throughout the

interactions with the user, in order to adapt to his or

her needs. Nowadays, with the multiplication of

mobile devices, people are connected all the time

and should be enabled to develop their learning

activities wherever they are and what device they are

using to.

Mobile learning (m-learning) is an emerging

concept as educators had initiated exploring mobile

technologies in teaching and learning environments,

and they cover from the ability to transmit learning

modules and administrative data wirelessly, to

enabling students to communicate with tutors and

peers “on-the-go” (Yuen and Yuen 2009). M-

learning refers to learning opportunities through the

use of mobile solutions and handhelds devices (i.e.,

PDA, mobile phones, smartphones and tablets) and

it is engaged to deliver right information and

resources – anytime and anywhere – to a specified

user, in a rich interaction, with powerful support for

effective learning, performance-based assessment

and strong search capabilities (Yuen and Yuen

2009). Presently, these mobile devices have become

cheaper and thus their usage by the students of our

universities is more common. In addition, in many

countries there is a government incentive for

lowering taxes for these devices for educational use,

offering a real possibility of use in high schools.

Our research in this paper goes further on the

functionalities and requirements of these systems in

order to be effective when learners develop their

learning activities using mobile devices. We are not

focusing just on mobile versions of e-learning

environment, which must be previously downloaded

and installed or specially developed for some group

of mobile devices. Indeed, we are focusing on the

real adoption of the m-learning environment

transparently by the student, adapted to student´s

context and needs.

To become this proposal a reality, is necessary to

enrich the learning objects (LOs) available to each

395

Gasparini I., Marilza Pernas A., S. Pimenta M., Palazzo M. de Oliveira J., Kemczinski A. and Geraldo H. Cavalheiro G..

m-AdaptWeb

 : AN ADAPTIVE E-LEARNING ENVIRONMENT FACING MOBILITY - Adaptation and Recomendation Processes based on Context.

DOI: 10.5220/0003921303950400

In Proceedings of the 4th International Conference on Computer Supported Education (CSEDU-2012), pages 395-400

ISBN: 978-989-8565-07-5

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

activity in the e-learning environment, providing

different versions to be chosen in running time,

according to student’s location, device, interests,

behavior, learning activity, knowledge and

performance as being presented in the m-learning

environment. Also, we present our integration of two

fields of study, recommender systems and adaptive

systems - in a synergistic way. Our approach deals

with creating context-aware recommendations of

different LO and filtering and adapting the user

interface, the navigation and the LOs taking in

account the user´s context and situation as well.

This work is structured as follows. Section 2

presents a background on e-learning environments

(ELEs). Section 3 present general concepts of

adaptation and recommendation and theirs

connection with ELEs. In section 4 we show the

infrastructure of m-AdaptWeb. A case study focused

on the internal operation of m-AdaptWeb is

presented in section 5, with our proposal for adapt

the student’s interface and recommend learning

content. The work finishes in section 6, with our

conclusions and future works.

2 E-LEARNING ENVIROMENTS

Many are the ELE projected to manage distance

learning. They are constantly improved in order to

better fit the exigencies of real persons facing real

learning problems. Some examples of these systems

are: Moodle (Moodle, 2011), BlackBoard

(BlackBoard, 2011) e SAKAI (Sakai, 2011).

The main objective of those environments is to

structure and manage content, not necessarily

enabling: content adaptation; student-teacher

interaction facilities; and authoring tools. These

facilities are not found in traditional e-learning

environments, but they have been required as much

as these systems are more intensively used and

easily accessible via Web.

AHS have the ability to adapt and personalize the

systems content, navigation and presentation, and

can incorporate some recommendations to each

student. This means that the systems must be able to

anticipate the needs to users and provide them with

recommendations of items that they might

appreciate based on their interaction with the system

and with other user. Focusing on e-learning

environments, this implies in personalize the

interface and navigation to student’s, helping them

facilitating their learning, and recommending the

best LO, adapted by the student´s needs, tasks,

profile and context.

There are many AHS works described in the

literature, like for example De Bra (2008) and

Canales et al. (2007). In our research the

experiments are developed into the AHS

AdaptWeb

, which consists on an adaptive web-

based learning environment developed by the efforts

of different Brazilian academic institutions.

AdaptWeb

is an e-learning environment able to

adapt hypermedia courseware contents and

navigation to student’s characteristics and

preferences. It was developed in PHP language and

handles a MySQL data base to store the students and

learning data. The software is free and available at

the sourceforge website (AdaptWeb, 2003)

The adaptive character of AdaptWeb

is mainly

supported by the structuring phase of a discipline

(e.g. Introduction to Programming, Artificial

Intelligence or Calculus). To develop the discipline

content, the author registers all concepts and

materials related to each topic. After that,

AdaptWeb

generates XML that represent the

domain content of each discipline in particular, and

is used by adaptation module to filter the different

learning objects that are linked to a student profile.

Storing files in XML format makes possible to

structure data in a hierarchical way, making possible

to filter the content of a discipline to determine

which content have to be present to each learner and

how. In this sense, discipline content may be applied

to different courses (e.g. Computer Science,

Mathematics, Physics), adapting which concepts and

their related documents have to be presented.

2.1 Managing Learning Objects

A learning object (LO) is defined as any entity,

digital or non-digital, that may be used for learning,

education or training (IEEE, 2002). The LOs have

several characteristics which justify their use. Ferlin

et al. (2010) describe the characteristics

differentiating them in technical and pedagogical.

The technical characteristics are related to the

standardization, storage, transmission and reuse of

LOs. Among these features, stand out: reusability,

interoperability, granularity, classification and

adaptability. The pedagogical features focus on the

construction of knowledge from the use of LOs and

on the concern in their construction. These features

are: interaction, autonomy, cooperation, cognition

and affect (Kemczinski et al., 2011).

Seeking to provide solutions for storing,

managing and searching LOs, a lot of repositories

were developed like: MERLOT (2008), LabVirt

(2010), BIOE (2010), OE³/e-tools (2010) and

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation

396

Interred (2010). In this work, we adopted the

repository ROAI (Learning Object Repository for

Computer Science and Informatics) (Kemczinski et

al., 2011) to store the LOs. The set of metadata

adopted in this work are presented in subsection 4.1.

3 RECOMMENDING PROCESS

AND ADAPTABILITY

Recommender system describes any system that

produces individualized recommendations as output

or has the effect of guiding the user to interesting or

useful objects in a large space of possible options

(Burke, 2002). Recommender systems can apply the

following kinds of techniques in recommendation

(Jannach et al., 2011): i) Collaborative

recommendation, implies implicit or explicit

collaboration among users; ii) Content-based

recommendation, based on the availability of item

descriptions and a profile that assigns importance to

these characteristics; iii) Knowledge-based

recommendation, when historical information is not

available, this technique makes use of additional

information (like technical feature of products) for

recommendation; and iv) Hybrid approaches, this

approach combines different techniques to generate

better or more precise recommendations.

In the special case of e-learning environments,

the recommendation process has to take into account

student’s relevant information, like student’s

interests, goals and problems. In our research,

beyond the user modelling and the learning objects

information, we propose the use of student´s context

and the experienced mobile situation in order to

guide the recommendation. To us, the student´s

context is considered as a 'broad notion' of relevant

data that can be used for recommendation.

The areas of recommender systems and AHS can

cooperate and actually one can benefits from the

other. However, we recognize that one significant

distinction relies in what is the final objective of

each one. Normally recommender systems provide a

set of resources that can be (or not) choose by each

user/student. By the other hand, AHS - besides

preparing different recommendations - also prepare

and execute different adaptations and

personalization for each user/student. Our work

integrates both approaches in a synergistic way: our

system is able not only to 1) creating context-aware

recommendations of different LO and the students

could choose accept or not, but also to 2) filtering

and adapting the user interface, the navigation and

the LOs taking in account the user´s context and

situation as well. In the next section we describe the

main characteristics of our system.

4 INFRASTRUCTURE OF

m-AdaptWeb

In this section we present our infrastructure for

modelling the user context in order to support

recommendations and a richer adaptability focused

on mobile devices. As presented in Figure 1, the m-

AdaptWeb infrastructure has to deal with data stored

in database repositories, related to: the LOs;

technological and location data; the student profile,

preferences and culture; learning domain data, which

represents the taxonomy of terms being.

First, in the LO repository, we have the different

LOs defined by the teachers and available to the

students. Each LO is defined according to a specific

set of metadata content, aiming at describing its

different characteristics in order to be adapted to

different mobile situations experienced by the

students. Details about the metadata that describe the

LOs are presented in the subsection 4.1.

The infrastructure also contains the description

of the mobile devices, used by learners when

navigating into the environment, and the location of

this device. This data consist on sensor data, which

describes: the kind of device student´s is using; the

screen resolution; kind of navigation browser; the IP

address and the network speed. These features are

very important for the recommendation because the

environment cannot recommend some LO that is not

the most adequate for the used device.

The Learning domain provides a taxonomy of

terms related to the learning domain being taught to

the student. The “subject” element from the LOs

metadata is related with some specific item from the

learning domain. We need information about the

student profile, preferences and culture to

adapt/recommend some specific material. Thus, the

Student personal data storages all needed

information about the students including their

profiles and contexts. These data are request by

different ontologies in the semantic layer.

Our semantic layer includes two ontologies: the

Technological ontology and the Learning ontology,

which are described in details in (Pernas et al.,

2011a), and an Student ontology, that, extended

typical characteristics of the students, to include new

dimensions (e.g. personal, educational and cultural

context dimensions) and each dimension is

m-AdaptWeb®:ANADAPTIVEE-LEARNINGENVIRONMENTFACINGMOBILITY-Adaptationand

RecomendationProcessesbasedonContext

397

Figure 1: Infrastructure of m-AdaptWeb.

represented by a specific ontology that can be used

jointly or separated, depending on the information

that we have available. More information about each

dimension can be found in Gasparini et al. (2011)

and about the structure see Gasparini et al (2012).

The infrastructure presented in Figure 1 collects

all relevant information from the repositories and

merge it in order to define the current mobile

situation of the student, depending on the LO being

accessed; the technological requirements of the

mobile device and the student’s profile and context.

To this merge, we are using the ontology network

and situation ontology defined in (Pernas et al.,

2011a) from which is possible to represent the

relationships among the individuals, obtained from

the data repositories, and infer the mobile situation.

A practical application of the ontology network is

presented in (Pernas et al., 2011b), with the relevant

events and the learning situations detected.

The Recommendation and Adaptation engine is

responsible to filter the content and present the right

LO to the right situation detected. Details about the

engine are presented in our Case Study, in section 5.

4.1 Structure of the Learning Objects

As we are working with an already existent and

functional e-learning environment, but not designed

to provide adaptations required by mobile users, the

structure of the learning content had to be revised.

The diversity of LOs present in the web today,

and their relationship with the current e-learning

environment had led our group thought of how to

store and organize these different materials. In this

way we have been integrating the “Learning Object

Repository for Computer Science and Informatics

(ROAI)”, an online repository, available to students

in our university (Kemczinski et al., 2011) to

AdaptWeb

environment. As ROIA uses the Dublin

Core metadata standard, either adopted by us in our

Domain model, the communication with ROIA

repository was facilitated. The element properties

defined by Dublin Core and applied in the ROIA

repository are: Contributor, Coverage, Creator,

Date, Format, Identifier, Language, Publisher,

Relation, Rights, Source, Subject, Title and Type.

More details about them in (Dublin Core, 2011).

AdaptWeb

requests the LOs to the ROAI

repository, which returns the specified LOs. Diverse

elements of the Dublin Core standard (Dublin Core,

2011) are needed, for example, the Creator element

is important to AdaptWeb

to be able to check if the

professor is responsible to the LO. The Date helps

AdaptWeb

checks the date to the learning situation.

With the Format, AdaptWeb

searches this element

to compare to the learning style and culture-

awareness of the student. The Language helps in

discovering this element to recommend or not based

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation

398

on student´s skill in any language. The Relation is

used to recommend a different LO. The Subject and

Title are used to compare with the AdaptWeb

’

subject and title materials.

5 CASE STUDY

In the case study we describe two different learning

scenarios, where two distinct students, from

different regions of Brazil, are developing the same

learning activity. These two scenarios are focused on

students from the 8

grade of the middle school.

General scenario – learning activity: the students of

the middle school (8

grade) have to develop a

research work about the Liberal Revolutions occurred

in Brazil in the beginning of the XX century. They have

to choose one Revolution to explain and also relate the

chosen revolution with other revolutions that were

occurring in Europe at the same period.

Profile of student_1: José lives in Rio Grande do Sul, a

state in the south of Brazil. He is developing the

discipline of History, in the modality of distance

learning. As he is visually impaired, he needs special

functionalities when using a web-based learning

system. Also, his learning style is active (learn by

trying things out, enjoy working in groups) and verbal

(prefer written and spoken explanation).

José’s current situation: José is in the library,

searching for some material about the Farroupilha

Revolution. He is currently using his tablet to help him

in this activity, connected in the Wi-Fi network from his

school library.

Profile of student_2: Maria lives in Bahia, a state from

Northeast of Brazil. She is also doing the discipline of

History in the modality of distance learning. Her

learning style is active, but she is visual (prefer visual

representations of presented material, such as pictures,

diagrams, flows and charts). As she had lived for 2

years in Canada, she has skill in English and French.

Maria’s current situation: Maria is on the bus, going

to her college, using her smartphone. She is searching

for some material about the Sabinada Revolution,

another Liberal Revolution occurred in Brazil. She is

connected through a 3G data connection, which varies

according to her location.

Internal rules that represent the valid context of

José and Maria in the infrastructure:

Context of João = {

hasDevice (O

Student

.João,O

Device

.Tablet),

hasNationality (O

Student

.João,O

Nationality

.BR),

interac_pref (O

Student

.João, O

Student

.Style1),

hasKnowledge(O

Student

.João, O

Domain

.History),

hasCourse (O

Student

.João, O

Course

grade),

hasTask(O

Student

.João,O

Task

.RW_01),

hasRestriction(O

Student

.João, O

Restrct

.Visual),

hasLocation(O

Student

.João,O

Location

.IPLibrary),

style_1 (O

Student

.João, O

Student

.Verbal),

style_2 (O

Student

.João, O

Student

.Active),

hasConnection (O

Student

.João,O

Netwk

.High)}

Context of Maria = {

hasDevice (O

Student

.Maria,O

Device

.SmarthPhone),

hasNationality (O

Student

. Maria,O

Nationality

.BR),

interac_pref(O

Student

. Maria, O

Student

.Style2),

hasKnowledge(O

Student

. Maria, O

Domain

.History),

hasCourse (O

Student

. Maria, O

Course

grade),

hasTask(O

Student

.Maria,O

Task

.RW_01),

hasLocation(O

Student

.Maria, O

Location

.Transit),

style_1 (O

Student

. Maria, O

Student

.Visual),

style_2 (O

Student

. Maria, O

Student

.Active),

hasConnection (O

Student

. Maria,O

Netwk

.Medium),

langSkill (O

Student

.Maria, O

Student

.English),

langSkill (O

Student

.Maria, O

Student

.French)}

5.1 Recommendation and Adaptive

Interface generation

AdaptWeb

already allows a certain level of

adaptive navigation. However, with the architecture

improvements’, it is possible to provide another

ways of adaptation, being aware of a broad number

of student’s context and, consequently, adopt other

ways of adaptation and recommendation.

As presented in Figure 1, our infra-structure has

a recommendation and adaptation engine. In this

engine we have been using different adaptive

techniques. First, in the adaptation of interaction, we

use the two categories of adaptation: content

adaptation (De Bra, 2008) and link adaptation (De

Bra, 2008). In the content adaptation category we

use the removing fragments technique to eliminate

all elements in the interface that do not correspond

to the student´s device, nationality and interaction

preferences.

In the link adaptation category we combine two

techniques: adaptive link annotation is used to

shows students their current state in the learning

process, by differing links already known, in study,

available to learn but not yet clicked. Also in the

menu, could be links that students’ do not have

sufficient knowledge yet and for that we used the

adaptive link hiding technique (by disabling the link

and present only the name of the topic.

In order to make recommendations, we evaluate

the metadata of the LOs stored in the ROAI to filter

the list of LOs that suite the student´s context and

situation properties in the mobile situation detect in

the last process step. In the first filter we evaluate

the metadata related to the Subject of the LO (and

the student’ knowledge and course); the Format of

LO (if it is appropriated to the student´s restriction)

and the Coverage and Description of LOs to analyze

if this LOs fit the students’ task. The second filter is

m-AdaptWeb®:ANADAPTIVEE-LEARNINGENVIRONMENTFACINGMOBILITY-Adaptationand

RecomendationProcessesbasedonContext

399

responsible to evaluate the metadata related to

Relation and Type to rate which of them cover the

student´s learning styles; evaluate the Language

metadata; and finally we re-evaluate the metadata

related to Format of the LOs with respect to network

connection restrictions.

6 CONCLUSIONS AND FUTURE

WORKS

This work presented a practical view of the m-

AdaptWeb. m-AdaptWeb is a module developed to

provide situation and context-aware behaviour to

mobile students in courses of the e-learning

AdaptWeb

, an adaptive e-learning environment. A

detailed description of how this context-awareness is

put in practice is provided, including the

infrastructure, the extended student´s model and its

representation in complementary ontologies.

A simple but actual contribution of this work is

the description of one feasible infrastructure –

particularly containing a recommendation and

adaptation engine - for adapting the content and

navigation taking in account properties of student´s

context in a specific mobile situation. We think that

our work has also a potential contribution, trying to

provide some directions of how to apply mobile-

oriented contextual aspects in e-learning

environment design. Our intention is to provide a

basis to discuss practices that address mobility

dimensions and cultural-aware issues for helping

designers incorporating such aspects in their e-

learning design process.

ACKNOWLEDGEMENTS

This work has been partially supported by CNPq,

CT-Info/CNPq and by CAPES, Brazil.

REFERENCES

AdaptWeb Project Description, 2003. SourceForge.

Available at http://adaptweb.sourceforge.net

BIOE, 2010. Banco Internacional de Objetos

Educacionais. http://objetoseducacionais2.mec.gov.br/

BlackBoard website, 2011. http://www.blackboard.com/

Burke, R., 2002. Hybrid Recommender Systems: Survey

and Experiments. In User Modelling and User

Adapted Interaction. pp. 331–370.

Canales, A. et al., 2007. Adaptive and intelligent web

based education system: Towards an integral

architecture and framework. In International Journal

of Expert Systems with Applications, pp.1076–1089.

De Bra, P., 2008. Adaptive hypermedia. In: J.M.

Pawlowski, H.H. Adelsberger, D. Sampson & Kinshuk

(Eds.), Handbook on Information Technologies for

Education and Training (2

edition). pp. 29–46,

Heidelberg: Springer.

Ferlin, J. et al., 2010. Metadados Essenciais: Uma

Metodologia para Catalogação de Objetos de

Aprendizagem no Repositório Digital ROAI. In: XXX

Congresso da Sociedade Brasileira de Computação,

pp. 1147–1156.

Gasparini, I., Pimenta, M. S., Palazzo M. de Oliveira, J.,

2011. How to apply context-awareness in an adaptive

e-learning environment to improve personalization

capabilities?. In: XXX Int.Conference of the Chilean

Computer Science Society (SCCC-JCC), Curicó.

Gasparini, I., et al, 2012. Improving User Profiling for a

Richer Personalization: Modeling Context in E-

Learning. In S. Graf, F. Lin, Kinshuk, & R. McGreal

(Eds.), Intelligent and Adaptive Learning Systems:

Technology Enhanced Support for Learners and

Teachers, ch 12, pp. 182-197, IGI Global.

IEEE Learning Technology Standards Comitee, 2002.

Draft Standard for Learning Object Metadata, IEEE

1484.12.1-2002.

INTERRED. 2010. http://interred.cefetce.br/interred/.

Jannach, D., Zanker, M., Felfernig, A., Friedrich, G.,

2011. Recommender Systems: An Introduction.

Cambridge University Press, NY.

Kemczinski, A. et al., 2011. Repositório de Objetos de

Aprendizagem para a Área de Computação e

Informática-ROAI. XXII SBIE/XVII WIE, pp.234–243.

LABVIRT, 2010. Laboratório Didático Virtual. http://

www.labvirt.fe.usp.br/.

MERLOT, 2008. Multimedia Educational Resource for

Learning and Online Teaching. http://www.merlot.org/

Moodle, 2011. Website. http://moodle.org/

OE³/E-TOOLS, 2010. Objetos Educacionais para

Engenharia de Estruturas. http://www.cesec.ufpr.br/

etools/oe3/

Pernas, A. M., Diaz, A., Motz R., Palazzo, M. de Oliveira,

J., 2011a. Situations and Ontology Networks to Define

Adaptive Actions in E-Learning Systems. In IADIS

International Conference WWW/Internet, Rio de

Janeiro, pp. 237–244.

Pernas, A. M., Palazzo M de Oliveira, J., 2011b. Enabling

Situation-Aware Behavior in Web-Based Learning

Systems. In: XXX Int. Conference of the Chilean

Computer Science Society (SCCC-JCC), Curicó.

Sakai, 2011. Project website. http://sakaiproject.org/

The Dublin Core

Metadata Initiative, 2011. http://Dublin

core.org/

Yuen, S. C., Yuen, P. K. 2009. Mobile Learning: Learning

on the Go. In D. Taniar (Ed.), Mobile Computing:

Concepts, Methodologies, Tools, and Applications, ch.

1.10, pp. 108-116, IGI Global.

CSEDU2012-4thInternationalConferenceonComputerSupportedEducation

400