Integrating Cloud Services to Support the Formation of Informal

Learning Groups

Adam Giemza

, Marc Jansen

and H. Ulrich Hoppe

University of Duisburg-Essen, Duisburg, Germany

University of Applied Sciences Ruhr West, Bottrop, Germany

Keywords: Cloud Services, Mobile Application, Group Formation, Learner Support.

Abstract: Cloud Services like Google Drive, Brainstormer, Doodle are meanwhile frequently used as tools for learn-

ing in various contexts. They provide storage facilities, production tools and particularly also coordination

support. The management of these heterogeneous tools is a challenge for the individual users as well as for

the usage in groups. This paper presents a mobile application to support the learners in the formation of in-

formal learning groups and integrates heterogeneous cloud services to support group formation and further

group work in a campus environment. Furthermore, this paper tackles the ambiguity of the term cloud ap-

plication and tries to provide a definition and discusses the process of bringing the Brainstormer web appli-

cation into the cloud for integration with the mobile application.

1 INTRODUCTION

Nowadays presence-based university studies are

supported by a complex and heterogeneous infor-

mation infrastructure: Course-specific information is

typically embedded in learning management sys-

tems, with possibly different systems used even in

one university. Administrative and organizational

information about study programs, general student

services etc. is often provided through other infor-

mation channels. Students access, partially store and

further manage this information on personal devices

such as notebooks or smartphones. Generally availa-

ble social media and cloud services may be used to

share and further distribute such information. In this

sense, a modern campus information environment is

an example of a complex, heterogeneous and some-

what scattered infrastructure (cf. (Hanseth, 2010)).

From the given characterization we can directly

construe the challenge of better integration. Integra-

tion can be addressed from two sides: from the

sources (i.e. information and service providers) or

from the user (student) perspective. Integration from

the source would need to be based on strong premis-

es of being able to change large parts of grown sys-

tems. So, in a modest and realistic approach we

address the problem from the user perspective.

We have specialized the problem in the form of a

typical application scenario: Especially students in

their first semesters on campus face the challenge of

finding peer groups to collaboratively work on as-

signments or prepare for exams. This includes the

problem of group formation and basic support for

group work. As a further specification, we have set a

focus on delivering and managing information on

personal mobile devices and on considering oppor-

tunities in space and time in a campus environment.

This conceptual and technical challenge has been

addressed in a still ongoing software development

project with master level students from an interdis-

ciplinary study program on interactive media and

applied cognitive science.

2 RELATED WORK

Providing orientation support for freshmen on cam-

pus has been a theme of a number of mobile learning

applications (Lucke, 2011), (Giemza et al., 2012),

often also using a game-based approach. While these

approaches focus on learning the university campus

and the surroundings, our application focuses on

getting to know new fellow students to form learn-

ing groups. The gamification aspect has a lower

importance, as we believe that the students have an

724

Giemza A., Jansen M. and Ulrich Hoppe H. (2013).

Integrating Cloud Services to Support the Formation of Informal Learning Groups.

In Proceedings of the 5th International Conference on Computer Supported Education, pages 724-730

DOI: 10.5220/0004455307240730

 SciTePress

intrinsic motivation (Ryan and Deci, 2000) to find

learning groups for gaining knowledge from others

in the collaborative learning process.

Jansen et al. (Jansen et al., 2005) describe a pro-

totypical campus information system that integrates

interactive public displays with personal mobile

devices and supports personalized and location-

aware information transfer. If such an infrastructure

(including interactive displays distributed over the

campus) were widely available and accessible this

would have been an ideal basis for implementing the

specific functions supporting group formation and

support of group work. However, unfortunately we

cannot rely on such premises.

The formation of learning groups has been stud-

ied from an intelligent systems perspective using

quite sophisticated processing techniques (Largillier

and Vassileva, 2012), (Isotani et al., 2009), (Hoppe,

1995). In our case, we do not rely on deep

knowledge modeling and intelligent learner diagno-

sis as input to the group formation process, although

this could be considered in future versions. Our

starting point and current focus is on opportunistic

usage of simple user profiles, general information

and coincidences in time and space.

Jansen et al. (Jansen et al., 2013) have classified

the usage of existing cloud services from an educa-

tional perspective. They distinguish different types

of services, including archiving/repository services

(such as Google Drive), communication and coordi-

nation services (such as Twitter or Doodle) and rich

production services beyond simple text messages

(such as MindMeister). Still, simple repository or

communication services are predominant in many

educational applications. A future potential is seen

in the area of processing services and especially

services for learning analytics.

3 Meet2Learn

Meet2Learn aims at supporting students to meet new

fellow students with the goal to form learning

groups for their joint lectures. Once the learning

group has been formed, the system will act as an

organizer to integrate learning materials and other

artifacts worked out collaboratively from different

cloud services.

In Meet2Learn learning groups are informal

groupings organized by students themselves, in

contrast to exercise groups offered beside a lecture.

Here the learners take the initiative to organize and

execute the learning group. Nevertheless they are

formed around a lecture that builds the context and

the content of the learning group. To support learn-

ers in creating and finding appropriate learning

groups, the system needs to be aware of the lectures

being offered at the University. This will prevent the

users to create multiple learning groups concerning

one and the same lecture only due to different nam-

ing by the learner while searching for a group or

while creating it. Furthermore learners need to be

represented in the system to provide grouping sup-

port based on a user profile. This profile needs to

contain the course of studies, the number of semes-

ters and finally the attended lectures. This will em-

power the system to provide recommendations based

on user-based collaborative filtering methods (Res-

nick et al., 1994). Finally, universities provide vari-

ous learning management systems (LMS) like

Blackboard or Moodle, which can be integrated with

the learning groups to import learning material from

the lecture as well as to export collaboratively elabo-

rated content back into the LMS. This content will

be exported from the “production-type” cloud ser-

vices like Google Drive or Brainstormer (Collide

Brainstormer, 2013).

In the sequel we will explain the components of

the Meet2Learn applications in more detail. The

target groups for the first prototype of Meet2Learn

are students at the University of Duisburg-Essen

studying Applied Computer Science and Applied

Communication and Media Science. Therefore the

system imports the lectures from the university sys-

tem called LSF and exports learning groups into a

Moodle provided by the university as well.

3.1 Architecture Overview

The frontend of Meet2Learn has been developed as

a native mobile Android application. This results in

the benefit of a practically permanent availability to

the students for checking the status of learning

groups and receiving notifications about relevant

information. The backend uses an agent architecture

on top of a blackboard system (Weinbrenner et al.,

2007). This allows for a flexible and extendible

design with connection to heterogeneous services

using multiple programming languages. The integra-

tion of cloud services takes place on the server side

through agents as well as directly on the client side

through the Android application (see

Figure 1).

The application integrates the following types of

cloud-based services on the smartphone: Communi-

cation, Production and Repository. Twitter and

IntegratingCloudServicestoSupporttheFormationofInformalLearningGroups

725

Facebook are integrated as the first type of service.

Hereby students can announce learning groups to

their friends that are not yet using the Meet2Learn

application. This will increase not only the number

of users but also the probability that a learning group

will take place with enough participants. Further-

more it will increase the extrinsic motivation (Bro-

phy, 2004) to the learner, as he or she announced the

intention to participate in a learning group to the

public. Google Drive and Brainstormer are integrat-

ed as cloud-based production services. Here learners

produce learning artifacts collaborative and link

them to the learning group for sharing the results.

Dropbox is integrated as a cloud-based repository

service. Learners can link uploaded resources and

provide direct access through the learning group.

The client-based integration makes use of the rich

connectivity and extendibility of Android through

flexible built-in and third-party APIs.

Figure 1: Meet2Learn Architecture Overview.

The server side uses an agent architecture utilizing

the SQLSpaces (Weinbrenner et al., 2007) as a

shared memory. The agents have different tasks

assigned. In the lower left of

Figure 1 importing

agents gather data from different heterogeneous

sources and import them into the system. The cur-

rent configuration integrates the study program from

the central repository of the university (LSF). This

data is used to support the students in adding lec-

tures to their profiles and prevents multiple notations

for one lecture. The management agents are respon-

sible for processing data and for controlling the

system. One particular feature is the notification of

the learners about new events (e.g., a new learning

group of interest will take place closely to the users

current location). This is implemented by the native

Android notification mechanism using the Google

Cloud Messaging (GCM) service. Finally, exporting

agents integrate learning groups into external sys-

tems, like the aforementioned learning management

systems (LMS). We are planning to integrate the

university’s Moodle by exporting Meet2Learn learn-

ing groups into Moodle cohorts. This will allow

learners to continue the work of a well-established

learning group inside Moodle closely located to the

lecture.

3.2 Mobile Application

The main entry point for the learner to the system is

the Meet2Learn mobile application. Figure 2 shows

three main views for the user. The first view (a)

represents the main entry point to the application.

Here the user finds a “news feed” containing the

latest information about learning groups as well as

notifications with recommendations and reminders

to the learner. The messages are typed by a color tag

on the left and can be therefore quickly grasped and

identified. A click on the message opens a new view

presenting more details about the event. The second

screen (b) shows the group creation screen. The

learner can configure a new group by setting multi-

ple parameters like the name, time and place and

publish this new group to the system. Fellow stu-

dents can search this group and join it. Finally, the

users are responsible themselves to participate phys-

ically in the informal learning group.

The last screen (c) shows the details a learning

group. The learner can see the title, the participants,

the next meeting and finally the resources shared

within the learning group using the aforementioned

cloud services (here Dropbox and Google Drive).

For this purpose Meet2Learn registers itself as a

recipient of “share events” of other applications on

the Android phone. A user can therefore open the

Google Drive app and share the link to a resource

with the Meet2Learn app. This will prompt the user

to select the correct learning group to share the doc-

ument with. All participants of this group will re-

ceive a notification that a new resource has been

shared with them in the shared group. The view (c)

also implements the standard share button (top right)

that allows the users to publish a learning group on

cloud-based communication services like Twitter

and Facebook. In this example, the user can share

the information about the learning group for the

lecture “InfoN” by clicking the share button. The

system will present a list of possible (cloud) services

to use to process the message. If the user selects

Twitter, the system will open the common Twitter

dialog with a pre-filled message about the group and

CSEDU2013-5thInternationalConferenceonComputerSupportedEducation

726

Figure 2: Meet2Learn App - (a) News feed and notifica-

tions (b) Group creation view (c) Group details view.

a link to a web page presenting some more details

about the learning group. Then the user can send the

tweet and share this to his or her followers. This

allows promoting the learning group to other learn-

ers not using Meet2Learn and thus allows new users

to join the Meet2Learn community.

4 CLOUDIFICATION OF WEB

APPS: THE BRAINSTORMER

EXAMPLE

In the previous sections we have mentioned Brain-

stormer (Collide Brainstormer, 2013) as a rich pro-

duction cloud-based service for collaborative brain-

stormings on assignments. In fact, we have devel-

oped Brainstormer as a Web Application in first

place and later deployed to the Open PaaS (Platform

as a Service) provider Cloudfoundry

. We will use

this example and our experiences to discuss what

“cloudification” actually means beyond mere web

Cloudfoundry - http://www.cloudfoundry.com/ Last visited:

Feb. 2013

applications and what possible added values can be

gained.

There is still no distinct definition or description

of the requirements for Cloud Applications. There-

fore, this section provides at first our view of a pos-

sible definition of the notion of Cloud Application.

Based on this definition and on the Brainstormer

example, we provide a technical description of steps

that are necessary for moving an already existing

application into a Cloud Computing environment.

4.1 A Possible Definition for Cloud

Applications

Having a look at one of the major definitions of

Cloud Computing (Mell and Grace, 2009), a distinc-

tion in three different level of abstraction can be

found:

1. IaaS - Infrastructure as a Service: Virtual provi-

sion of computing power and/or memory. A

prominent example of an IaaS service is the Am-

azon WS service.

2. PaaS – Platform as a Service: Provision of a

runtime environment, like application servers,

databases etc. Cloudfoundry is an example for a

(open) PaaS provider.

3. SaaS – Software as a Service: Provision of usual-

ly browser based applications that can directly be

used. Here, Google Docs or the Customer Rela-

tionship Management software of salesforce.com

serves as examples.

Obviously, the only layer the end-user actually

interacts with, is the SaaS layer. Therefore, Cloud

Applications will be deployed to this level of ab-

straction. Nevertheless, in order to provide Cloud

Computing key factors to the Cloud Application,

usually, the SaaS layer will be deployed on top of a

PaaS layer running a flexible IaaS environment.

Therefore, the first thing to mention for a Cloud

Application is, that it needs to run in a Cloud Com-

puting environment, as explained above.

Furthermore, (Mell and Grace, 2009) provides a

list of essential characteristics an application needs

to provide in order for being called a Cloud Applica-

tion. First of all, the application needs to provide an

on-demand self-service, which basically means that

the necessary computational resources are provided

to the application without human interaction. In the

context of learning scenarios, this would mean that a

Cloud Application would allow the learner to use the

application independent of time and space, with

almost unlimited resources. The second aspect men-

IntegratingCloudServicestoSupporttheFormationofInformalLearningGroups

727

tioned is the broad network access, meaning that the

application itself can be accessed via standardized

tools and/or off-the-shelf applications (Pettersson

and Vogel, 2012). The aspect of resource pooling is

not that much of interest with respect to the SaaS

layer since the provision of according resources is

usually ensured by deploying the SaaS layer on top

of a PaaS and especially an IaaS layer. Therefore,

this characteristic does not provide any new impacts

for a Cloud Application within the context of learn-

ing scenarios. The fourth aspect, the rapid elasticity,

yields to the high degree of scalability that Cloud

Applications need to ensure. Here, with respect to

learning scenarios, this characteristic does not pro-

vide new requirements, since, similar to the aspect

of resource pooling, this aspect would be answered

by a deployment to a PaaS and/or a IaaS infrastruc-

ture. Taking a broader view to the question about

scalability issues for applications in the context of

learning scenarios (Giemza et al., 2012), the ques-

tion of scalability provides an interesting topic of

current research. The last characteristic mentioned is

the characteristic of service measuring. Here, for

Cloud Computing environments, a special need for

measuring the service consumption is necessary,

especially since the resource provided for the service

in question are highly flexible. Again, with respect

to learning scenarios, this characteristic is not of

high interest as it would usually also be handled on

the PaaS and IaaS level and/or there is no special

business model for learning applications so far.

Therefore, a Cloud Application in the context of

learning scenarios, might be defined as a highly

scalable application, both from a technical and a

collaboration point of view, that is available via

standardized interfaces and with almost unlimited

resources.

4.2 Necessary Steps for Moving

a Learning Application

to the Cloud

In order to fulfill the requirements discussed in the

last subsection, there are a couple of steps necessary

in order to extend an already existing application to

a Cloud Application. We will discuss these steps in

general and also in the context of the Brainstormer

example.

First of all, the accessibility of the application

needs to be evaluated. The overall process for the

modification of an already existing application to a

Cloud Application is shown in

Figure 3.

Here, the major question is, whether the applica-

tion is available to large audience (in our context the

learners) via standardized interfaces. Of course this

becomes pretty easy in case of web applications, as

with Brainstormer, but might be a little bit more

crucial with respect to non web-based applications.

Here, usually, the question is not only about the user

interface with whom the learner usually interacts,

but also the question of how data can be added (in

terms of input channels (Bollen et al., 2012)) to the

applications, might be an indicator for the availabil-

ity of the application itself. Brainstormer shows also,

that these questions are not mutually exclusive, as it

provides a web interface as well as an API for third-

party clients to contribute content.

The next would be to deploy the application to

an IaaS infrastructure. Depending on the architecture

of the application, this step might be more or less

trivial. Our experience from the deployment of two

learning applications to an IaaS infrastructure shows

that this step is usually not as trivial as it might ap-

pear on the first view. This layer is however already

covered when using PaaS providers as Cloudfound-

ry.

Figure 3: Necessary process for the modification of an

existing application to a Cloud Application.

Basically, the problems that usually occur can be

subdivided into two different categories. The first

category yields towards the data layer of the applica-

tion. Usually, applications used in learning scenarios

depend on some sort of database that stores the de-

veloped learning objects and provides additional

material to the learner. The examples described in

(Giemza et al., 2012) and (Bollen et al., 2012) that

we tried to deploy to an IaaS infrastructure, the type

of database that the application used before was not

supported. Therefore, we had to re-implement the

database layer. Of course another solution to this

CSEDU2013-5thInternationalConferenceonComputerSupportedEducation

728

challenge might have been to use our already exist-

ing database, but this would have resulted in the

question whether we can ensure the necessary scala-

bility, availability and performance of the database.

Especially these kinds of problems should be solved

by the deployment of the application to an IaaS

infrastructure. The case of Brainstormer was easier,

as Brainstormer uses a MongoDB database in the

backend. Cloudfoundry provides instances of Mon-

goDB, therefore the migration was limited to a re-

configuration of the database settings.

The second step for moving an application to a

Cloud Computing environment, takes the used

frameworks and external API’s into account. One of

the major characteristics of modern software devel-

opment is the high degree of re-use of already exist-

ing frameworks and API’s. Here, it is important to

ensure that the provider supports the frameworks

and API’s used for the development of an applica-

tion. The usage of standard external services, e.g.,

Web Services, demands special attention. It needs to

be ensured that not only the application itself is

highly scalable (from a technical point of view) but

also that the used external services are scalable in

the same dimension.

Finally we can conclude, that Brainstormer can

be considered as a Cloud Application with respect to

our definition and the explained steps to migrate a

Web Application to a Cloud Application.

5 CONCLUSIONS

AND OUTLOOK

The Meet2Learn application is an example of inte-

grating and managing information from heterogene-

ous sources to support personalized and group learn-

ing. It takes the heterogeneity of the surrounding

information infrastructure as a given and aims at

user-side integration. It combines general infor-

mation from the campus environment with personal

profiles and location information. It also includes the

use of several types of cloud services. In this con-

text, the previously developed Brainstormer applica-

tion for generating and collecting ideas in a group

has been “cloudified”.

The system will be ready for use in the summer

term (starting in April 2013). It will be evaluated

with a group of beginners from Interactive Media

and Applied Cognitive Science. From a systems

perspective, the application will be enhanced by

using ontologies and semantic processing for identi-

fying specific user interest and needs.

REFERENCES

Hanseth, O. (2010). From Systems and Tools to Networks

and Infrastructures - from Design to Cultivation: To-

wards a design Theory of Information Infrastructures.,

In Jonny Holmstrøm (ed.), Industrial Informatics de-

sign, Use and Innovation. IGI Global: 122-156.

Lucke, U. (2011). A pervasive game for freshmen to

explore their campus: Requirements and design issues.

in Proceedings of the IADIS International Conference

on Mobile Learning. Spain. pp. 151-158.

Giemza, A., Verheyen, P., & Hoppe, H. U. (2012). Chal-

lenges in Scaling Mobile Learning Applications: The

Example of Quizzer. 2012 IEEE Seventh International

Conference on Wireless, Mobile and Ubiquitous

Technology in Education. pp. 287–291

Ryan, R., & Deci, E. (2000). Intrinsic and Extrinsic Moti-

vations: Classic Definitions and New Directions. Con-

temporary Educational Psychology, 25(1), pp. 54-67

Jansen, M., Uzun, I., Hoppe, H.U., Rossmanith, P. (2005).

Integrating Heterogeneous Personal Devices with Pub-

lic Display-Based Information Services. Proceedings

of IEEE WMTE 2005: 149-153

Largillier, T. & Vassileva, J. (2012). Using Collective

Trust for Group Formation. Procedings of the 18

In-

ternational Conference4 on Collaboration and

Tehcnology (CRIWG 2012). Springer: 137-144.

Isotani, S., Inaba, A., Ikeda, M., Mizoguchi, R. An ontolo-

gy engineering ap-proach to the realization of theory-

driven group formation. International Journal of Com-

puter-Supported Collaborative Learning 4(4), 445-478

(2009)

Hoppe, H. U. The use of multiple student modeling to

parameterize group learning. In Proceedings of AIED

‘95. Washinton D.C., August 1995. 234-241.

Jansen, M., Bollen, L., Baloian, N., Hoppe, H.U. (2013).

Cloud Services for Learning Scenarios: Widening the

Perspective. In Proceedings of the 1st International

Workshop on Cloud Education Environments. Guate-

mala. pp. 33-37

Resnick, P., Iacovou, N., Suchak, M., Bergstrom, P., &

Riedl, J. (1994). GroupLens: An Open Architecture for

Collaborative Filtering of Netnews. In R. Furuta & C.

M. Neuwirth (Eds.). Proceedings of the 1994 ACM

conference on Computer supported cooperative work.

pp. 175-186

Collide Brainstormer – http://brainstormer.cloudfoundry

.com/ - Last visited: Feb. 2013

Weinbrenner, S., Giemza, A., & Hoppe, H. U. (2007).

Engineering Heterogeneous Distributed Learning En-

vironments Using Tuple Spaces as an Architectural

Platform. Seventh IEEE International Conference

on Advanced Learning Technologies. pp. 434-436

Brophy, J. (2004). Motivating students to learn. British

Journal of Educational Technology (Vol. 36, p. 418).

IntegratingCloudServicestoSupporttheFormationofInformalLearningGroups

729

Lawrence Erlbaum Associates.

Mell, P., Grance, T. (2009). The NIST definition of Cloud

Computing. Version 15

Pettersson, O. and Vogel, B. (2012). Reusability and

Interoperability in Mobile Learning: A Study of Cur-

rent Practices. 2012 Workshop on Scalability and In-

teroperability Dimensions for Mobile Learning Appli-

cations (SID-ML 2012), In Proceedings of the Seventh

IEEE WMUTE 2012.

Giemza, A., Bollen, L., Jansen, M., Hoppe, H.U. (2012).

An Architecture for Supporting Heterogeneous Multi-

Device Learning Environments. In: Proceedings of the

7th IEEE International Conference on Wireless, Mo-

bile and Ubiquitous Technologies in Education

Bollen, L., Eimler, S.C., Jansen, M., Engler, J, (2012).

Enabling and Evaluating Mobile Learning Scenarios

with Multiple Input Channels. In: Proceedings of the

18th International Conference on Collaboration and

Technology (CRIWG)

CSEDU2013-5thInternationalConferenceonComputerSupportedEducation

730