ABEL-GUI: AN AGENT-BASED GRAPHICAL USER INTERFACE

FOR E-LEARNING

Steffen Kernchen, Fritz Zbrog and Reiner R. Dumke

Department of Computer Science, Otto-von-Guericke-University of Magdeburg, Magdeburg, Germany

Keywords:

e-Learning, Agent Technology, Graphical User Interface.

Abstract:

In this paper we present an agent-based graphical user interface (GUI) for the domain of e-Learning: the

ABEL-GUI. It is developed following a structured approach for the analysis, speciﬁcation and design of agent-

based graphical user interfaces. Therefore a logical goal/task hierarchy is designed and depicted by a society

of agents. By this an implicit partition of the GUI takes shape and gets connected with the particular agents.

The result takes beneﬁts from the cooperation idea behind agent technology and focuses on specialization,

personalization and interaction support.

1 INTRODUCTION

We think creating interfaces based on agent techno-

logy does not only follow the new paradigm of agent-

based software development but leads to several ad-

vantages, too. O’Malley et al. list several guide-

lines whether an agent-oriented approach is useful

(O’Malley and DeLoach, 2002). According to them

agent-based graphical user interfaces can be neces-

sary, because:

• GUI’s may include complex/diverse types of in-

teraction between components as well as to ex-

tern distributed heterogeneous resources. This

ﬂexibility of protocols can be easily supported by

agent technology.

• Negotiation, cooperation and competition may

occur among different entities.

• Some aspects of a GUI can have autonomous

characteristics.

• A modiﬁcation or expansion of the system can be

anticipated. By using agents this ﬂexibility is ex-

tended to modiﬁcation or expansion during run-

time of the system.

By choosing an agent-based approach additional

aspects of ﬂexibility can be applied for the domain

of e-Learning. An increased support of adaptation

for example leads to an improved adaptation to the

user’s needs. By choosing the developmentmethodol-

ogy presented in chapter 4 in combination with agent

technology level-based as well as agent-based person-

alization of the functionality and the appearance of

the GUI is possible. Basis of the adaptation are de-

mands, properties, preferences, skills or interactions

of the user that may occur or are explicitly modelled

within a user model.

By means of agent technology, a framework de-

velopment becomes possible. It can provide basic

functions to fulﬁl minimal requirements. Additional,

specialized agents can be added during runtime to ex-

tend the GUI. Thereby lightweight software provision

and functional adaptation are distinguishing features

of the system.

But why so much ﬂexibility for e-Learning? Lear-

ning is one of the most important aspects of our cul-

ture. Its effective implementation and support is es-

sential for human beings in all phases of their life.

Life-long learning is the corresponding slogan. There

are many changes in that long period of time. That

refers to internal human being aspects as well as to

external environmental ones. Human lifelong devel-

opment requires highly ﬂexible mechanisms for the

adaptation of supporting systems. That applies to

learning in a particular manner. External reasons are

new scientiﬁc knowledge about didactics, psychol-

491

Kernchen S., Zbrog F. and R. Dumke R. (2007).

ABEL-GUI: AN AGENT-BASED GRAPHICAL USER INTERFACE FOR E-LEARNING.

In Proceedings of the Third International Conference on Web Information Systems and Technologies - Society, e-Business and e-Government /

e-Learning, pages 491-494

DOI: 10.5220/0001271504910494

 SciTePress

ogy, interface design and new information about the

topic to be learned. Versatile problems require adap-

tive and extendable solutions.

Following an agent-oriented methodology for de-

signing graphical user interfaces we developed a

framework of a front-end for a highly adaptive e-

Learning system.

This paper is organized as follows. In section 2 we

present basic key aspects that need to be considered in

the context of our approach. The important factor of

usability as well as the plug-in technology and the ad-

ditional value contained in our approach are sketched

out. The used methodology for the development of

agent-based graphical user interfaces is presented in

section 3. Key elements and novel features of the de-

veloped ABEL-GUI are part of section 4. Section 5

closes this paper with conclusions and future work.

2 FUNDAMENTALS

Following Shneiderman and Plaisant user interface

design should be driven by user tasks and interface

mechanisms (Shneiderman and Plaisant, 2005). The

identiﬁed tasks must be real, complete and represen-

tative and the design that is developed following this

task-centred approach must be validated to be appro-

priate. The other key element is the communication

between user and system. The interface mechanisms

must provide appropriate support for the user to solve

his actual task.

There are three paradigms of user interface de-

sign described in (Tsou and Buttonﬁeld, 1998). Tra-

ditional user interfaces operate within a centralized

environment. They are not platform- and application

independent. Client/Server user interfaces are clients

remote connected with one server where data and pro-

gram are located. They are usually platform indepen-

dent like global user interfaces, which are not limited

to connect to only one server. These interfaces can

access multiple heterogeneous systems.

Traditional User Interface Client/Server User Interface Global User Interface

Interface

Programs

Data

Server

Client Interfaces

Client with Global

Interfaces

Distributed Servers

Figure 1: Three paradigms of user interface design (Tsou

and Buttonﬁeld, 1998).

Plug-in technology is an approach that targets the

concept of mobile codes. Plug-ins are supplement

software modules which are integrated in other soft-

ware. They are only working in order with the appli-

cation they are written for. Technologically they can

be implemented in various ways: .NET, DCOM and

Enterprise Java Beans to name a few.

Both approaches differ in the ﬂexibility of usage

and adaptation. Autonomous, pro-active and cooper-

ative characteristics of agent technology are aspects

that are not part of plug-ins, but provide essential ad-

vantages for the usage in context of e-Learning sys-

tems.

3 DEVELOPING AGENT-BASED

GRAPHICAL USER

INTERFACES

For developing agent-based graphical user interfaces

an iterative goal-directed methodology including de-

sign patterns and creative techniques is presented

here. As visualized in ﬁgure 2 there are several de-

sign stages.

Identification of goals

Hierarchical subdivision of goals

- depending on complexity

Identification of functional requirements

- definition of additional properties

Hierarchical subdivision of functions

- depending on complexity

Derivation of functional dependencies

- to other agents

- to extern components

- definition of additional properties

Determination of required and optional data

Goals

Tasks

Inter-

actions

Agent-based meaning GUI-based meaning

Presentation

containers

Visual and

functional

elements

Functional

dependences

Subdivision of functional dependencies

- depending on complexity

Post refinement activities

- mapping of funtional objects to agents

- detailed description of egdes and nodes

Figure 2: Stages of developing agent-based GUI’s.

At the ﬁrst stage the goals that should be solved

are deﬁned and reﬁned. The starting point is the given

problem in a special application domain. The goals

and sub-goals have general character and form a tree.

No special functions will be deﬁned yet.

WEBIST 2007 - International Conference on Web Information Systems and Technologies

492

The special functionalities that are needed to reach

the deﬁned goals are identiﬁed at the second stage. If

a function is too complex for an easy implementation

its subdivision in smaller nodes is recommended. The

complexity does not only refer to the functionality but

also to the amount of dependencies to other nodes or

external components.

A tree-shaped structure based on the subdivision

of goals and tasks is deﬁned so far. Now this structure

is extended by the deﬁnition of additional internal and

external connections. Next to the connections of the

speciﬁed goal and task hierarchy - the edges of the

tree - more internal functional dependenciesmay exist

and need to be speciﬁed: a graph structure develops.

Up to this point the general structure of the multi-

agent system and the graphical user interface is mod-

elled. The next steps are the mapping of functional

objects, respectively sets of those objects, to agents

and the detailed description of edges and nodes. Com-

plexity measures like the number of sub-functional

objects or the number of edges can indicate an appro-

priate mapping. Sometimes aspects of the later im-

plementation need to be taken into account, too. By

recombining functionalities all sub-functional objects

as well as all internal and external connections are

compressed into one agent without disregarding the

detailed functional design.

4 AGENT-BASED GUI FOR

E-LEARNING

As already stated out ﬂexibility, adaptation and per-

sonalization are key features of a GUI for e-Learning.

Using the previously described methodology we de-

veloped a graphical user interface based on agent

technology. By turning into account its heteroge-

neous and autonomous characteristics the intended

goals can be achieved.

ManagementLearning

RLO

User model

E-learning platform

Performance

observation

RLO database

User model server

Personal space Interaction

Communication

server

Figure 3: Core components of the ABEL-GUI and selected

transitive connections to external distributed elements.

The framework beneﬁts from the autonomy char-

acteristic of agent technology in terms of improved

modularity, adaptivity and ﬂexibility. Proactivity and

reactivity may directly result in adaptive changes

of the component presentation. An example is the

preference-based sortation of collaboratives in a chat

list.

In this section we introduce main concepts of the

architecture of the developed system from a student’s

point of view. Other roles are tutor, author and ad-

ministrator and lead to a distinct usage of existing el-

ements as well as to completely new features.

4.1 Learning Components

The ﬁrst and obviously most important parts of the

GUI to be developed are the learning components.

That refers to all goals and functions that are directly

coupled with the learning tasks.

Following the already described methodology we

identiﬁed four direct sub-goals and one direct func-

tionality, namely the course, course structure, domain

visualization, support and actual announcements.

Course

structure

Course

Domain

visualization

Support

Actual

announcements

RLO

User model

E-learning platform

Course

management

Performance

observation

RLO database User model server

Domain

ontology

Motivation

HelpAvatar

RSS-Feed

Hierarchy Node

Figure 4: Graph presenting learning components, sub-

components and functional edges.

4.2 Management Components

The adaptive functionality of this system is based on

technical as well as organizational management. That

refers to the management of agents, courses, user

model and to logging mechanisms.

Course

management

Agent

management

User model

E-learning platform

User model server

User model

management

System

management

Logging

Agent

catalogue

Visualization

Student

management

Student

observation

Student

tests

Figure 5: Graph presenting management components, cho-

sen sub-components and functional edges.

4.3 Personal Components

Personal parts that might be implemented in an e-

Learning GUI are e.g. a personal homepage, a course

scheduler,personal space for storing of ﬁles and infor-

mation, individualized search mechanisms and a kind

ABEL-GUI: AN AGENT-BASED GRAPHICAL USER INTERFACE FOR E-LEARNING

493

Course

Personal

homepage

User model

E-learning platform

User model server

Personal space

Individual

mechanisms

Diary

User server

Figure 6: Graph presenting personal components, sub-

components and functional edges.

of diary. A user server might be a technological base

for these elements.

4.4 Interaction Components

Human interaction is essential for learning and

learning-related motivation (e.g.: (Walther, 1992),

(Dimitrova et al., 2003)). Our approach offers vari-

ous core synchronous and asynchronous communica-

tion supporting elements and the easy usage, update

and extension of this set. Examples are a wiki, chats,

forums, user search component, etc.

All interaction components imply functional con-

nections to the user model, the user server and the

communication server. Preferences and logﬁles need

to be stored for every communication tool for person-

alization and adaptation.

users

Wiki

User

model

E-learning platform

User model server

ForumChat

User

Communication server

Text Video Audio

Figure 7: Graph presenting interaction components, chosen

sub-components and functional edges.

4.5 Implementation Remarks

This GUI is intended to be implemented as a frame-

work. That is an important aspect for the provision of

the ﬂexibility, the necessity and advantages of which

was proven in the introduction. Therefore minimal

required elements as well as necessary infrastructural

aspects must be deﬁned.

At this point we argue that the core elements

shown in ﬁgure 4, their direct sub-elements and all

elements, that are directly or transitive connected by

required edges, are non-optional parts of the frame-

work. This is visualised in the particular ﬁgures of

the other subsections of this section.

Infrastructural aspects that must be respected are

the identiﬁcation of agents, docking mechanisms,

interaction interfaces, the communication itself and

warranty of role-based access and modiﬁcation is-

sues.

Interaction is one of the most important features of

agent technology. Most of their advantages are based

on communication. In the context of interfaces exter-

nal and internal communication needs to be optimized

to guarantee the ﬂow of usage. Those are contrary

goals that must be carefully balanced and that directly

refer to the implementation technology.

Performing post-reﬁnement activities is a stage of

the used methodology that is between reﬁnement of

goals/functionalities and the implementation. It pools

sub-functions belonging together into one agent.

5 CONCLUSIONS AND FURTHER

WORK

In this paper we presented main concepts of the

ABEL-GUI, a graphical user interface for the do-

main of e-Learning, which was developed following

a graph-based approach for developing agent-based

GUI’s. High ﬂexibility, organized interaction mech-

anisms as well as adaptation support are key beneﬁts

of this technique. To examine the usefulness of our

approach we modelled the presented agent-based in-

terface for the e-Learning domain as a framework. An

integral part of our future research is its implemen-

tation using existing web technologies by extending

them with agent characteristics.

REFERENCES

Dimitrova, M., Sadler, C., Hatzipanagos, S., and Murphy,

A. (2003). Addressing learner diversity by promoting

ﬂexibility in e-learning environments. In Proceedings

of the 14th International Workshop on Database and

Expert Systems Applications (DEXA03).

O’Malley, S. A. and DeLoach, S. A. (2002). Determing

When to Use an Agent-Oriented Software Engineering

Paradigm, pages 188–205. Springer.

Shneiderman, B. and Plaisant, C. (2005). Designing

the User Interface – Strategies for Effective Human-

Computer Interaction. Pearson Education, 4th edition.

Tsou, M.-H. and Buttonﬁeld, B. (1998). An Agent-Based,

Global User Interface for Distributed Geographic In-

formation Services. In Proceedings of the 7th Inter-

national Symposium on Spatial Data Handling, pages

602–612.

Walther, J. B. (1992). Interpersonal effects in computer-

mediated interaction: A relational perspective. Com-

munication Research, pages 52–90.

WEBIST 2007 - International Conference on Web Information Systems and Technologies

494