ADAPTED AND CONTEXTUAL INFORMATION IN MEDICAL

INFORMATION SYSTEMS

Karine Abbas, André Flory

LIRIS-UMR 5205 CNRS, INSA de Lyon

, Claude Bernard Lyon1 University, Lumière Lyon 2 University

Ecole Centrale de Lyon, Bât.Blaise Pascal, 7 avenue Jean Capelle, 69621 Villeurbanne Cedex, France

Christine Verdier

LIG- UMR 5217 CNRS, team SIGMA, Joseph Fourier University

681 rue de la Passerelle, BP. 72, 38402 Saint Martin d'Hères Cedex, France

Keywords: Context, Profile model, Personalization, Information System, User-System Interaction.

Abstract: Today, due to the integration of the pervasive computing to current information systems, new challenges are

involved: the amount of data increases tremendously over time and users are more and more heterogeneous

with different needs and roles. These challenges require systems with some adaptation capacities. The

personalization and user modelling process are the key elements to propose solutions to these challenges.

For that, this paper presents personalized access technique which takes into account the different

requirements of personalization for different user’s needs and different contextual situations.

1 INTRODUCTION

Current information systems (IS) try to improve

continuously the access to data in different domains:

economic, scientific, technical, etc. However, the

considerable amount of data, the large spread of

mobile devices (Barkhuus & Day, 2003) providing

access to information anywhere and at anytime and

the heterogeneity of users with different needs have

finding relevant information and reducing data

overload quite challenging tasks. These different

challenges require systems with some adaptation

capacities. Systems should be capable of adapting

informations to user’s needs and his environment in

which he is acting (Abbas et al., 2006).

Commonly, several works are developed in the

literature in order to provide help to user when

research information. (Ceri et al., 2005; Fiala et al.,

2003). The majority of these systems are addressed

the issue of personalization by presenting suitable

models based on the user (Amato & Staraccia, 1999;

Razmerita, 2005). These user models consists

mainly in classifying the user characteristics into a

set of categories such as demographic attributes

(identity, personal data), professional attributes and

behaviour attributes (traces of navigation on Web),

goals, preferences, skills, accessibility, activity,

interest. The majority of these works is actually

specific solutions suited only for predefined

adaptation requirements and are hardly reusable for

a new adaptive application. Moreover these

techniques have focused on the user and rarely

considered the context in which a user interacts with

the system while user’s context can be a key element

to improve personalized information access process.

On the other hand, the context is commonly

considered in context-awareness systems. Theses

systems limit the notion of context, referring to the

situation in which the user is acting, to the concepts

of user’s location and device (Fogarty et al., 2004;

Lemlouma & Layaida, 2004; Chaari et al., 2007).

So, there is a need to have a system covering all

facets of personalization process. we propose a

personalized data access approach based on profile

management which take into account all facets

characterising the user and his context of use. This

approach requires two steps: the former consists in

building profile model characterizing a context (for

instance profile for user, role, device, etc.). The

latter uses the profiles content for personalizing the

exchanged data with the user.

363

Abbas K., Flory A. and Verdier C. (2008).

ADAPTED AND CONTEXTUAL INFORMATION IN MEDICAL INFORMATION SYSTEMS.

In Proceedings of the Tenth International Conference on Enterprise Information Systems - HCI, pages 363-366

DOI: 10.5220/0001700903630366

 SciTePress

Context_parameter Context Profile

This article is organized as follows: Section 2

presents the general architecture of the

personalization approach. Section 3 describes a

profile model, a context model and services model.

Section 4 shows how to use these profiles in the

personalization process.

2 GENERAL ARCHITECTURE

The general framework we propose requires

three notions:

 The user which uses the system in order to

access information;

 The resources extracted from databases in

response to the user’s query;

 A software module which manages the

personalization process based on profiles

modelling. To exploit these profiles, a set

of services is proposed for defining rules

which will be executed for personalized

information.

The software module is organized into two

layers:

 A conceptual layer which proposes

interconnected models: a profile model, a

context model and a service model. This

layer provides a meta-description of these

models.

 A application layer which consists in

instantiating these models and using it for

personalizing information access. We

distinguish three modules: a profile

management, a service management and a

exploiting profiles and services in the

personalization process.

Figure 1: Composition of software module.

The personalization approach can be used

independently for new applications as well as for

legacy systems which involve requirements of

personalization possibly not defined in advance.

3 PROFILE MANAGEMENT

Before descrinbing the profile modele, we start by

defining the context.

3.1 Context Modelling

In general, a context is defined by a set of external

parameters (Chaari et al., 2007). An instance of

these parameters characterizes a contextual situation.

A change of a parameter’s value defines a new

contextual situation. For example, we define two

contextual situations C1: {user = Smith, role =

doctor, location= office} and C2: {user = Smith,

role=doctor, location = patient at home}. In C1,

Doctor Smith may access a health database from her

office for screening patients for prevention care,

while in C2, the same doctor accesses the same

database at the patient home for post treatment

analysis. Then in different contextual situations,

users may access different data and exploit different

aspects of an application.

Figure 2: Description of context.

To cover all facets characterizing a context parameter,

a profile can be proposed. The profile modelling is

described in the section below.

3.2 Profile Modelling

In general, a profile is defined as a set of knowledge

characterizing a context parameter such as a user, a

role, a device, etc. To guarantee a generic and

evaluative model, we present a high abstract

description of profile.

Formalery, a profile is structured in a set of

primitives such as:

 An owner, for who/what the profile is

defined. The owner is a parameter of

context such as user, role, etc.

 A profile describes the information

characterized an owner. This information is

structured and classified into a set of

dimensions. Then a profile is related to one

or more dimensions.

 A dimension represents the information that

Context model

Profile model

Service model

Conce

tual la

Profile management

Services management

Application layer

Exploiting of the personalization process

ICEIS 2008 - International Conference on Enterprise Information Systems

364

characterizes a profile such as skills,

preferences, etc. A dimension is represented

by one or more attributes.

 An complex attribute describes information

on a dimension. This attribute is composed

on other (simple or complex) attributes

 A simple attribute is associated with a

value. The simple attribute can depending

on an concept which designantes a context

parameter or a classe of data sources of an

application.

Figure 3: Meta-description of profile.

The instanciation of the profile modele allows to

define the concrete elements of profile for a context

parameter (user, role, etc.). these profiles are

implemented in the XML documents.

Figure 4: A XML profile description.

Figure 4 illustrates an example of profile. The

owner is user. The profile is composed in two

dimensions: access right and preferences. The first

dimension is composed on a complex attribute

authorized and the second dimension is composed

on the complex attribute activity.

3.3 Service Management

To guarantee a generic personalization process,

services are created and associated to elements of

profiles (dimension or attribute). A service is

defined as an autonomous program. while a new

dimension (or complex attribute) is added, a new

service is created and associated with it.

A service is modelled by a set of input

parameters PInput={e

, e2, …, e

} and a set of

output parameters POutput={s1, s2, .., s

}, phase of

execution, criterion of execution. Then a service is

modelled as follows:

Service = nameS(E, S, criterion, phase)

The variable Criterion designates the condition

needed for execution of a service. We distinguish

three criterions:

 Sequential: This criterion specifies that this

service will be executed independently of

other services;

 order: this criterion specifies that this

service depends on other service.

 conditional: this criterion specifies that the

execution of this service depends on a

condition.

The criterion is defined by two attributes : (name,

value) as follows :

- Name={sequential, ordre, conditional} and value

 If name = sequential, then value = “;”

 If name = ordre then value =

nameS1

→nameS2

 If name = condition then valeur = "if"

<condition> "then" operation ["else"

operation] "end if" ;

A service is composed on a set of functions.

Then, when a service is invoked, these functions will

be executed.

Figure 5: Services organized in two phases.

The personalization process is applied on xml

data extracted from various databases in response to

user query. The process requires two phases (as

illustrated in figure 5). At each phase, a set of

Dimension

Name

Complex_att

Profile

Simple_att

concept

Owner

Name

describ

Parameter conte

Other class

Content

Value

Associate

1.*

1 1

<XML>

<Users>

</access right>

<activity> <location>hospital</location>

</preferences>

</profile>

</user>

<!other users>

</Users>

XML data XML

data

PhaseI: Content Personalisation Phase II :

resentation

Ser

ice

{

}

Service

{

…

ADAPTED AND CONTEXTUAL INFORMATION IN MEDICAL INFORMATION SYSTEMS

365

services are executed. The first phase is concerned

by the content personalisation and the second phase

concern to adapt the data presentation at the user

interface.

4 USING PROFILES FOR

PERSONALIZATION PROCESS

The personalization approach requires two steps:

The former selects the profiles which correspond to

user and his current context from profiles database.

The latter uses the profiles for personalizing XML

data by executing services associated to the profiles.

Figure 6: Different steps of personalization process.

The first step of personalization process consists in

selecting the profiles that are valid with regard to the

user’s current context. This selection is performed

by comparing the available profiles with the

contextual situation.

Once the profiles have been selected, the second

step consists in finding the services associated with

elements of the found profiles. For that, we extract

the structure of each profile. Then, for each element

of a structure, we search the service associated with

it from the services database.

Once the services are found, we must separate

them in two groups: a group which belong to phase I

(content personalisation) and a other group for

services belonged to phase II (presentation

personalisation).

Finally, services of phase I are executed followed

by services of phase II.

5 CONCLUSIONS

In this paper, we have proposed a personalization

approach based on the profile modeling for

information systems. The approach takes into

account different context situations in order to cover

all facets of personalisation needs.

Actually, we are planning to implement this

approach in the I2MS project (Sassi et al., 2007) in

order to adapt the user interface.

REFERENCES

Abbas, K., Verdier, C., 2006. Design of a medical

database transformation algorithm. International

Conference on Enterprise Information Systems -

ICEIS 2006, 23 – 27, Paphos – Cyprus.

Amato, G., Staraccia, U., 1999. User profile modeling and

applications to digital librairies, Proceedings of the

3rd European Conference on Research and avanced

technology for digital libraries, p. 184-187.

Barkhuus, L., Dey, A., 2003. Is Context-aware computing

taking control away from user? Tree levels of

interactivity examined. In: Proceeding of Ubicomp

2003, Springer, pp. 150-156

Burell, J., Gray, G.K., Kubo, K., Farina, N., 2002.

Context-aware computing: a text case. In Borriello, G.,

Holmquist, L.E. (eds): 4

Int. Conf. on Ubiquitous

Computing, LNCS 2498. Springer, 34-38.

Ceri, S., Daniel, F., Demald, V., Facca, F.M. 2005. An

approach to user-Behavior-Aware Web Applications.

In 5

Int. Conference on wen Enginnering (ICWE),

Springer-verlag, Australia.

Chaari, T., Ejigu, D., Laforest, F., Scuturici, V..M., 2007.

A Comprehensive Approach to model and use Context

for adapting Applications in Pervasive Environments.

Int. Journal of Systems and Software, Elsevier.

Fiala, Z., Hinz, M., Meissner, K., Wehner, F., 2003. A

component based approach for adaptive dynamic web

documents. Journal of web Engineering, 2(1-2):53-73

Fogarty, J., Lai, J., Christensen, J., 2004. Presence versus

availability: the design and evaluation of a context-

aware communication client, Int. J. Hum. Compt. Stud.

61 (3) , pp. 299-317

Lemlouma, T., Layaida, N., 2004. Context-aware

Adaptation for Mobile Devises. IEEE Int. Conf. on

Mobile Data Management, IEEE Computer Society

106-111.

Razmerita, L., 2005. User modeling and personalization

of the Knowledge Management Systems, book chapter,

in Adaptable and Adaptive Hypermedia, to be

published by Idea Group Publishing.

Sassi, S., Verdier, C., Flory, A., 2007. A new system for

project management: the I2MS interface. in TIGERA,

Tunisie Hammamet.

Profiles database

Contextual Situation

Selection of profiles

Profiles

Step 1: Selection of

profiles according to

contextual situations

Services

management

Exploiting of profiles

and services

User interface

document.xml in

response to user query

Step 2: using profiles

in the personalization

process

ICEIS 2008 - International Conference on Enterprise Information Systems

366