AN ADAPTABLE MIDDLEWARE FOR PERSONALIZING WEB

APPLICATIONS

Zahi Jarir

1,2

And Mohammed Erradi

Cadi Ayyad

University, Faculty of Sciences Semlalia Marrakech

Department of Computer Science

B.P. 2390, Bvd. Prince My Abdellah, Marrakech, Morocco

Computer Networks and Multimedia Research Group , UFR-RT

Mohamed V Souissi University, ENSIAS

B.P. 713, Agdal, Rabat, Morocco.

Keywords: Web Personalization, Dynamic Reconfiguration, Computational Reflection, Reflective Middleware,

Separation of concerns, Component-based Application, EJB, JOnAS, Web application.

Abstract: The personalization is an important topic for the Web industry. It consists in providing the capabilities to

accommodate Web applications to user’s requirements such as defining preferences on the execution of the

application, associating the provided application to a specific terminal, specifying or modifying QoS

parameters, and so on. The contribution of this paper is to present a solution to ensure an advanced Web

application personalization by focusing on the middleware level rather than the application level. We

provide an enhanced architecture to personalize Web applications using the EJB technology. An

implementation using JOnAS environment is presented. It has the advantage to adapt and/or reconfigure

Web application’s behavior at runtime according to the user’s specific needs.

1 INTRODUCTION

Web applications users become more and more

experienced and thereafter require more advanced

and sophisticated Web applications. In fact these

“smart users” are interested in Web applications that

are easy to reconfigure and/or adapt to their new

requirements. The users may require defining

preferences related to the application’s execution

such as fluctuating network bandwidth, and/or

associating the Web application to a specific

terminal, and so on. For instance certain users may

wish to adapt their Video-On-Demand web

application on their PDA instead of their laptop by

selecting voice rather than sub-titles. The capability

to accommodate applications to the user’s

preferences such as technological restrictions,

mobility requirements, resources constraints, etc. is

called personalization.

The personalization may consist in offering the

possibility to adapt dynamically or statically Web

applications to build ad-hoc applications related to

the user’s needs. In the new provision environments,

Web applications Providers are looking to introduce

new architectures and techniques to allow users to

have an active role and to be involved in the

personalization process of their applications.

Personalizing Web applications is therefore a

challenging task and becomes even more powerful

when applied to advanced and multimedia

applications. Consequently, new tools and

architectures for personalization, especially dynamic

personalization, are required. In this direction a new

paradigm, called engineering adaptive Web

applications, have been emerged.

The rest of this paper is organized as follow. In

Section 2, we discuss related work. In Section 3, we

argue the motivations to choose EJB platform and

thereafter we expose the main entities of the EJB

architecture. In Section 4, we show how the EJB

architecture can be extended to allow dynamic

adaptability of services. In Section 5, we apply the

suggested approach to personalize a Web application

which consists in interacting with a Multimedia

Conferencing System before concluding in Section6.

287

Jarir Z. and Mohammed Erradi A. (2005).

AN ADAPTABLE MIDDLEWARE FOR PERSONALIZING WEB APPLICATIONS.

In Proceedings of the First International Conference on Web Information Systems and Technologies, pages 287-292

DOI: 10.5220/0001234602870292

 SciTePress

2 RELATED WORK

Several approaches in adaptive Web applications are

currently investigated, using different techniques

and methodologies, in different areas e.g.

eCommerce, Distributed eLearning Environments,

Telecommunication Services, etc. These approaches

concern mainly the following topics:

a. Web Information Personalization that focuses

on searching and browsing the Web information

using personalized Web search systems. Recently

researchers are interesting to define how to enable

personalization functionality to personalize the

interaction with web content (Henze et al., 2004;

Shahabi, 2003).

b. Web Presentation Personalization that

contributes to tailor the information presented and

the structure to the user’s preferences, knowledge

or interests. This topic focuses on how adaptation

can be implemented through the manipulation of

links, and content and presentation of nodes

(Fortier et al., 2001; Goderis et al., 2001; Koch et

al., 2002). One sub-category of this context is

terminal adaptivity (Hinz et al., 2004).

c. Web Application Personalization, that

contributes to tailor the Web application structure

and/or behavior to satisfy specific user

requirements. To address this kind of

personalization there is mainly two ways :

o Using conceptual approaches that consist

in modeling and designing personalized Web

applications. The aim of this approach is to

bring dynamic personalization support in

Web conceptual modeling constructs by

creating new conceptual modeling approach

or extending the existing ones e.g. OOH,

OOHDM, etc. (Garrigos et al., 2003).

o Using implementation approaches that

consist in modifying the implementation

details of the developed Web applications.

These approaches introduce personalization

features using different techniques such as

mobile agent, reflection techniques, and so on

(Maknavicius et al., 1999; Jarir et al., 2002).

Most of the evoked researches focus mainly to a

specific Web application concern such as building

tailored presentation interface, showing personalized

information, adapting application's functionalities,

etc. However less attention has been paid to build an

architecture that covers almost all the concerns of

Web personalization.

Building an architecture for personalizing Web

application imply to deal with different

personalization concerns that must be seamlessly

integrated. That’s why we have elaborated as a first

step a deep taxonomy showing the varieties of

personalization concerns. Based on this taxonomy, a

new problem emerges: what is the best level among

the application level and middleware level to deal

with the diversity of personalization concerns

(diversity of users devices, evolvable execution

context, etc.)?

Recall that Web applications are usually

developed over a Middleware which in turn is

implemented over a protocol stack to ensure data

communication and to manage the communication

resources. Performing dynamic changes to Web

applications become a tedious task and need to

consider changes at the underlying levels especially

at the middleware level. The idea behind this

approach is that applications built on the top of the

middleware rely on it for all interactions with the

execution environment, so adapting the middleware

allows us to indirectly adapt the applications.

Moreover at the middleware level, the Web

application may be adapted to a given configuration,

either by replacing some components of code or by

adding new components in order to enhance existing

functionalities. Furthermore it is very difficult to

adapt Web application to the continuous variations

of their execution contexts at the application level.

3 WHY ENTERPRISE JAVABEANS

ARCHITECTURE

Object-Oriented Middleware technologies such as

CORBA or Java RMI have proved their suitability

for standard client-server applications. However,

such platforms do not provide the required levels of

adaptation and/or reconfiguration that are needed to

accommodate the diversity of modern distributed

applications including support for multimedia, real

time, mobility, etc. This motivates many middleware

research groups to built new and advanced

middleware technologies.

Sun Microsystem, OMG and Microsoft are

aware of these limitations that’s why the current

developments of the Enterprise Java Beans (EJB),

CORBA Components Model and COM+ are

proposed. The main focus of these platforms is to

alleviate the application-level programmability

issues by hiding, within the so-called component

containers, a large part of the complexity with more

declarative interfaces.

The growing popularity of EJB architecture is

due to the advantages offered to the distributed and

Web-based applications, e.g. faster application

development, ability to build complex applications,

separation of business logic from presentation logic,

WEBIST 2005 - WEB INTERFACES AND APPLICATIONS

288

application interoperability, etc. Furthermore,

currently there are more than 30 implementations

(free and commercial) of EJB servers and that

number is increasing (Emmerich et al., 2002).

Therefore, we choose the EJB technology as an

example of component-based middleware to prove

that Web application personalization can be dealt

with at the middleware layer.

Enterprise JavaBeans (EJB) is a specification

and architecture for the development and

deployment of distributed server-side, transactional,

and secure business application components.

Therefore EJB servers reduce the complexity of

developing middleware by providing automatic

support for middleware services such as

transactions, security, database connectivity, and

more. The EJB specification defines also several

standard roles and responsibilities that introduce

three fundamental. These entities are EJB server,

EJB Container and EJB component.

However, the EJB platform does not address the

needs for adaptation and extension required in

several applications. Our suggested solution focuses

on the EJB architecture to show how it can be

extended to allow reconfigurations and extension at

runtime providing an adaptable EJB infrastructure

(Jarir et al., 2002). This research was supported by

the RNTL project ARCAD (ARCAD, 2005).

4 TOWARDS AN ADAPTABLE EJB

INFRASTRUCTURE

In this section we first show how to introduce

runtime adaptability in the EJB environment. Then

we present our approach for an adaptable EJB

infrastructure. This infrastructure is made using the

computational reflection (Maes, 1987), which is a

suitable technique for dynamic adaptation features.

4.1 How to make EJB adaptable?

According to the separation of concerns paradigm

used by the EJB architecture, the code of an EJB

application is split in two parts: the functional code,

representing EJB components, and the non-

functional code, representing middleware services

(e.g. transactions, persistence, security, etc.).

However, the configuration between EJB

components and Middleware services is only

supported at deployment-time using a declarative

deployment descriptor. This descriptor, presented in

XML format, defines a set of accessor methods for

setting and getting information about the Entreprise

Java Beans being deployed.

To make EJB adaptable we need to make

explicit the separation between functional code and

non-functional code. Then, the associations between

these two kinds of codes can be modified at runtime

according to the user’s requirements, to personalize

its application. This reconfiguration is made by

identifying which piece of the functional code is

affected (and how it is affected) by the non-

functional code.

To ensure an advanced adaptability of an EJB

application to the desired changes, we need to focus

on the EJB container layer. This is because the EJB

container is an intermediary between the EJB

components and the outside world especially

between the EJB component and the access to

various resources and EJB services. The containers

are generated statically using the information

provided by the EJB component’s deployment

descriptors. These descriptors cannot modify the

associations between the EJB components and the

EJB services at run-time without modifying the

containers.

4.2 Computational reflection

Computational reflection, or just reflection, has been

proposed as a solution to the problem of creating

applications able to maintain, use, and change

representation of their own designs. It is defined as

the capability to monitor and modify dynamically

the structure and the behavior of a system. A

reflective system is therefore able to use self-

representations to extend, modify and analyze its

own computation (cf. Figure 1).

Program

Interpreter

Data

Problem

Domain

Reflective System

Represent

Compute

Figure 1: Reflective system

AN ADAPTABLE MIDDLEWARE FOR PERSONALIZING WEB APPLICATIONS

289

4.3 The adaptable EJB infrastructure

JOnAS (Java Open Application Server) (Jonas,

2005) is an Open Source implementation of the

J2EE

specification. It is a pure Java

implementation of this specification that relies on

the JDK. It is part of the ObjectWeb Open Source

initiative. The Opening of JOnAS environment

opens us the way for introducing the reflection

features to try to make EJB architecture adaptable.

In addition, JonAS environment offers an open

source tool, called the GenIC (Generate Interposition

Classes) that allows generating the EJB container

code. Therefore this tool will guarantee to set up our

approach by focusing on the EJB container layer as

mentioned earlier.

In order to respect the EJB container

specification, we have delegated the task of a

dynamic composition of services to another object

called DynamicComposite, representing the meta-

object of the EJB container as

shown in Figure 2.

The set up of this indirection is made thanks to the

computational reflective features that allow diverting

all methods call from EJB container to its associated

DynamicComposite object. This object is able to

compose dynamically attached or detached EJB

services before or after sending the method call to

the EJB component. Therefore the

DynamicComposite object will be responsible for

playing the role of a dynamic composer of the EJB

services.

To perform the required reconfiguration

(attachment and/or detachment of EJB services) for

each application, a generic adaptation engine is

introduced. This reconfiguration is guided by the

adaptation policies, described in XML format, and

they are initiated by the engine when significant

evolutions of the environment are detected by a

simple monitoring framework, consisting of a

collection of probes (CPU usage, battery life,

bandwidth measure…).

These adaptation policies are:

- System policies, consisting in sets of rules of the

form condition ⇒ action, where the condition

is related to the execution environment (as

reified by the monitoring framework), and the

action is either the attachment or detachment of

a specific services, possibly with configuration

parameters.

- Application policies, which define groups of

EJB components according to their runtime

properties and bind existing system policies to

these groups.

The following XML code presents an example of

a system policy named “bandwidth-policy”, which

are interpreted at run-time by the adaptation engine.

<system-policy name="bandwidth-policy">

<rule>

<when>

<less-than>

<property-value name="/system/network.bandwidth"/>

</less-than>

</when>

<parameter name=”SoundEncoder” property-

value=”classLpc”/>

</updated>

</ensure>

</rule>

</system-policy>

EJB Server

Physical Resources

System.xml

Application.xml

EJB Services

Transaction

Dynamic

Adaptation

EJB Container

Enterprise

JavaBean

EJBObject

Home Object

DynamicComposite

…

Monitoring

Framework

Adaptation

Engine

Database

Service 3

Service 2

Service 1

Figure 2: An adaptable EJB infrastructure

WEBIST 2005 - WEB INTERFACES AND APPLICATIONS

290

5 AN ENHANCED EJB

ARCHITECTURE FOR WEB

APPLICATIONS

PERSONALIZATION

The increasing popularity of the World Wide Web

and the diversity of hardware appliances make the

Internet users crave to have the privilege to control

the functionalities of the Web applications to which

they are subscribed. For example a user may need to

obtain a MPEG2 Format for a movie, obtained from

a Video-on-Demand server, instead of a classic VHS

quality received by default.

The Web application personalization considers

many types of reconfiguration. These

reconfigurations may concern changes of:

- Parameters of the application that concerns the

application’s data, as for example QoS

parameters of the multimedia applications (e.g.

debit, video resolution, etc.)

- Functional aspect of the application that

concerns the application’s behavior such as:

o Adaptability that affects the application’s

behavior without calling new components or

functionalities within the application. This

personalization consists in activating and/or

deactivating some of their already existing

functionalities.

o Extensibility that corresponds to introduce

new additional behavior or functionalities in

the application to answer a specific need.

- Technological aspect that concerns the

modification of the application in order to be

executed on different platforms (e.g. operating

system, etc.), different types of terminals, the

run-time variations of availability of certain

resources such as CPU, memory, communication

capacities, and so on.

To build personalizable Web application, a more

flexible architecture, that takes into account those

varieties of adaptation and/or reconfiguration, is

needed.

To build a more flexible architecture for Web

applications personalization, we have reused the

adaptable JOnAS EJB infrastructure that allows

component-based applications to be aware of, and

adapt to, the variations in the execution context.

Therefore our goal was to see how this infrastructure

could be transformed in response to all varieties of

reconfiguration and/or adaptation studied before.

Figure 3 shows the elaborated architecture for

Web applications personalization. This EJB-based

architecture has the advantage to be able to ensure a

runtime tailorability of the Web applications

behavior. This tailorability concerns both the

behaviors of each EJB service and also each other’s

composition that build the desired Web application.

More precisely, the dynamic personalization takes

into account the attachment and/or detachment of a

service and also the update of a specific attached

service behavior. This personalization can be

activated according to:

• The expressed user’s requirements on the

Graphical User Interface (GUI). These requirements

are thereafter translated to a specialized policy,

called personalization policy, that will be performed

by the adaptation engine incorporated inside the

Dynamic Adaptation service. After the required

actions expressed by this policy are handled by the

DynamicAdaption Service, the EJB Client will

thereafter modify the GUI presentation according to

these changes. The following code shows an

example of a personalization policy:

<personalization-policy>

<parameters name=”resolution” value

="1024x768"/>

<parameters name=”VideoEncoder” value

="MPEG1"/>

</updated>

Physical

Resources

Monitoring

Framework

EJB

Client

EJB Server

System.xml

EJB Services

Transaction

Dynamic

Adaptation

EJB Container

Enterprise

JavaBean

EJBObject

Home Object

DynamicComposite

…

Adaptation

Engine

Database

Video

Service

Chat

Service

Audio

Service 1

Personalization.xml

Figure 3: An EJB architecture for Web applications personalization

AN ADAPTABLE MIDDLEWARE FOR PERSONALIZING WEB APPLICATIONS

291

<parameters name=”AudioEncoder” value

="classGSM"/>

</attached>

</personalization-policy>

• The variations of the client environment that

considers the technological restrictions, the

continuous variations of the execution

environment, etc. This detection is handled by

the monitoring framework, placed at the client-

side, in response to the conditions introduced by

the system policies rules. To allow the EJB

client to be informed about what type of

variations can be taken into account to notify the

EJBObject, we have injected implicitly the

conditions of the system policies rules during

the initialization of the application.

6 CONCLUSION AND

PERSPECTIVES

The contribution of this paper deals with Web

applications personalization. Specifically, we have

shown how the personalization can be dealt at the

Middleware level by developing an adaptable

JOnAS EJB infrastructure. This infrastructure

consists in modifying at run-time and with a fine

granularity the association between EJB components

and middleware services. Based on this EJB

infrastructure, we have presented an adaptable EJB

architecture to personalize Web applications. This

enhanced architecture allows users to take an active

role in personalizing dynamically their Web

applications, and also enables Web applications to

be aware of, and to adapt to, the variations in the

execution environment.

Currently we are interested in the extensibility of

the Web application by adding a required service

from other application service provider. However

this extensibility may cause deviation from desired

behavior or systems failure. This issue generates a

broad topic, named feature interaction problem that

requires service (or feature) interference detection

and resolution.

REFERENCES

ARCAD, http://arcad.essi.fr

Emmerich, W., & Kaveh, N., 2002. Component

Technologies: Java Beans, COM, CORBA, RMI, EJB

and the CORBA Component Model. Proc. of the 24th

Int. Conference on Software Engineering, Orlando,

Florida. pp. 691-692. ACM Press.

Fortier, A., Rossi, G., & Cappi, J., 2001. Using Meta-

Level Techniques to Personalize O-O Applications”,

Object-Oriented Programming, Systems, Languages,

and Applications OOPSLA 2001/ECOOSE.

Garrigos, I., Gomez, J., & Cachero, C., 2003. Modelling

Dynamic Personalization in Web Applications. In

Proc. 3 rd Int. Conf. Web Engineering, volume 2722

of LNCS., pages 472–475. Springer Verlag,

Goderis, S., Rossi, G., Fortier, A., Cappi, & J., Schwabe

D., 2001. Combining Meta-level and Logic-Based

Constructs in Web Personalization, In Proc.of the 6th

International Computer Science Conference on Active

Media Technology, Pages: 57 – 64.

Henze, N., & Kriesell, M., 2004. Personalization

Functionality for the Semantic Web: Architectural

Outline and First Sample Implementations. First

International Workshop on Engineering the Adaptive

Web.

Hinz, M., Fiala, Z., Wehner, F., 2004. Personalization-

based Optimization of Web Interfaces for Mobile

Devices; In: Proceedings of 6th International

Conference on Human Computer Interaction with

Mobile Devices and Services, Glasgow, Scotland.

Jarir, Z ., David, P.C. & Ledoux, T., 2002. Dynamic

Adaptability of services in Entreprise JavaBeans

Architecture , Seventh International Workshop on

Component-Oriented Programming /ECOOP 2002,

Malaga, Spain.

Jarir, Z., & Erradi, M., 2002. Telecommunication Services

Customisation , Association for the Advancement of

Modelling and Simulation Techniques in Enterprises,

Vol. 23 n°3, pp. 1-14, AMSE-Journal.

JOnAS, http://www.evidian.com/jonas

Koch, N. et al, 2002. Patterns for Adaptive Web

Applications. In 7th European Conference on Pattern

Languages of Programs, Irsee, Germany.

Maes, P., 1987. Concepts and Experiments in

Computational Reflection , in Proc. Conference on

Object-Oriented Programming systems, Languages

and Applications, Orlando, FA, pp. 147-155.

Maknavicius, L., Koscielny, G. & Znaty, S., 1999.

Customizing Telecommunication Services : Patterns,

Issues and Models, , vol. 1597, pp. 194-209, Lectures

Notes in Computer Science Springer-Verlag.

Shahabi, C., 2003. Web Information Personalization:

Challenges and Approaches, In the 3nd International

Workshop on Databases in Networked Information

Systems , Aizu-Wakamatsu, Japan.

WEBIST 2005 - WEB INTERFACES AND APPLICATIONS

292