WWW2003 Poster #261

Five Models for Integrating a Progressive Access Approach into Web-based Information Systems

Marlne Villanova-Oliver
laboratory LSR-IMAG
BP 72, 38402 St Martin dHres
Grenoble, FRANCE
(+33) 4 76 82 72 80
Marlene.Villanova@imag.fr
Jrme Gensel
laboratory LSR-IMAG
BP 72, 38402 St Martin dHres
Grenoble, FRANCE
(+33) 4 76 82 72 80
Jerome.Gensel@imag.fr
Herv Martin
University Joseph Fourier
BP53, 38041 Grenoble cedex
Grenoble, FRANCE
(+33) 4 76 82 20 90
Herve.Martin@ujf-grenoble.fr

ABSTRACT

Web-based Information Systems (WIS) are used for processing and diffusing a large amount of information over the Internet. It is therefore crucial to adapt both the content and the presentation of information to users in order to save them from some disorientation or cognitive overload syndromes. For this purpose, we introduce the notion of Progressive Access which aims at giving WIS users a flexible and personalized access to data, by stratifying their information space. These stratifications are described through a Progressive Access Model which is connected with four other models: a Data Model, a Functional Model, a Hypermedia Model and a User model. We explain how these five models integrate the progressive access approach in the design of WIS.

Keywords

Web-based Information Systems, design, adaptability, progressive access

1. INTRODUCTION

Web-based Information Systems (WIS) are Web applications that allow to collect, structure, store, manage and diffuse information, like traditional Information Systems do, but over a Web infrastructure. Generally, WIS have to deal with large amounts of information, with distinct sources and formats, and are characterized by complex services they offer to manage information. Then, WIS design methods have to cope with the multiple facets of a WIS, namely the content (data), the functionalities (services) and the hypermedia features (structure, navigation and presentation).

Relying on the Web and, thus, on a hypermedia structure, a WIS allows its users to navigate through a wide information space. However, this benefit can turn out to be a drawback for users, causing both a disorientation syndrome and a cognitive overload. The disorientation syndrome (also called the "lost in hyperspace syndrome") is experienced by users who browse an information space which have a complex hypermedia structure, and then get lost in it. Also, users can be subject to a cognitive overload when they have to face a too massive and difficult to understand quantity of information. These two drawbacks are highly prejudicial to the life of the WIS. For this reason, information must be organized, managed, and displayed in a personalized way and so that users are efficiently given access to pertinent information. WIS adaptability becomes then a crucial issue. It can be defined as the ability a WIS has to provide its users with relevant information with regard to their rights, needs, individual characteristics and material configurations, etc.

The central idea behind this work [1] is that the user of a WIS does not need to access all the information all the time. Our objective is then to build a WIS which has the capacity to give access to its resources (i.e. information and services) gradually and in an adapted way. First, information/services considered as essential for a user are provided, and then, some complementary information/services, if needed, are proposed through a guided navigation. We propose a generic model, called the Progressive Access Model (PAM), in charge of organizing information/services and of providing a support for the progressive access. In our approach, the PAM is moreover closely related to four other models: the Data Model, a Functional Model, the Hypermedia Model and the User Model. Together with some methodological guidelines, these five models allow designers to define WIS able to provide theirs users with a personalized and progressive access to both information and services, which contributes to limit the disorientation syndrome and the cognitive overload.

2. PROGRESSIVE ACCESS MODEL

Basically, the progressive access relies on the following notions. We call stratification the process (and its result) which aims at defining a particular organization of a Maskable Entity (ME) into an ordered set of Representations of Maskable Entity (RoME). A ME is a set of at least two pieces of information. A RoME is a subset of pieces of information defined from a ME. Each of the RoMEs which, considered as a whole, constitute the stratification of a given ME, is associated with one level of detail. Moreover, the RoME defined at level i is included in the RoME defined at level i+1 (strict inclusion). Thus, accessing to a RoME means accessing to a more (resp. less) detailed representation of the ME through a navigation mechanism called unmasking function (resp. masking function).

The Progressive Access Model (PAM), expressed in UML, gives the rules for specifying well-formed stratifications. Figure 1 shows the main stereotyped classes of the PAM and associations between them. An instance of the class Maskable Entity is, at least, composed of two instances of the class Element of Maskable Entity. An instance of the class Stratification is represented as an aggregation of, at least, two ordered instances of the class Representation of Maskable Entity. An instance of this class is linked by the association adds to one or more instance(s) of the class Element of Maskable Entity which are the elements of the ME added by the RoMEi at the level of detail i. The dependency relation ({subset}) ensures that the added elements belong to the set of elements corresponding to the ME. The ternary association definition (which links the classes Stratification, Maskable Entity and User Category) allows the designer to specify as many stratifications as needed for a given Maskable Entity. This way, each user can benefit from a personalized progressive access.


Figure 1. The Progressive Access Model described in UML

3. DESIGN OF PROGRESSIVE ACCESS

We propose to combine the PAM with four other models in order to take into account the progressive access approach in WIS design:

  1. The Data Model describes the entities of the application domain targeted by the WIS. By applying the principles of the PAM to this Data Model, the designer can define the appropriate stratifications for the entities of the application domain.

  2. The Functionality Domain describes and organizes the tasks users can perform within the WIS. The organization relies on three levels supported by concepts of different granularity: the functional space, the functional role and the functionality. The functionalities are identified from UML use cases where actors are persons using the system. At an intermediate level, the functionalities defined for a same actor are grouped together in a functional role. At the highest level, a functional space, defined for every user, gathers the various functional roles a user has at her/his disposal. Each of these three concepts can be considered as a Maskable Entity and, consequently, be referenced by the PAM in order to offer a progressive access at each of the three levels.

  3. The Hypermedia Model is composed of two sub-models: a Structuration and Navigation Model and a Presentation Model. The former describes the structure of the hypermedia in terms of nodes and links of navigation. Nodes and links are both related to progressive access concepts. Basically, nodes correspond to the stratification, ME and RoME concepts, links to the masking and unmasking functions. The hypermedia structure is derived from the stratifications performed on the Functional Model concepts as explained above. Hence, the Presentation Model is dedicated to the composition and the appearance of the Web pages. We use composition and graphical charters to ensure the hypermedia pages coherence and to manage the appearance of concepts related to progressive access.

  4. The User Model can be seen as an extension of the PAM which has to be attached to the class User Category. Such an extension integrates some characteristics for representing users. In our approach, we propose a progressive access oriented extension which distinguishes between Group and User. A group of users gathers information about users who play the same functional role. A user belongs to one or more groups. This model contains references about the stratifications established for each group (resp. user).

These four models have been defined in order to integrate the principles of the progressive access into each of the dimensions (content, functionalities, hypermedia features) of a WIS and, moreover, considering adaptation to user. To reach this goal, we widely exploit the possibilities offered by the PAM, the reference model in our approach. Figure 2 shows the relations that hold between this model and the four other ones.


Figure 2. Links between the four design models and the PAM

4. CONCLUSION AND FUTURE WORK

We have defined and presented the notion of Progressive Access as a contribution to adaptability in Web-based Information Systems (WIS). This notion helps in limiting the cognitive overload and the disorientation a user may feel when, browsing in a WIS, she/he is confronted to too much information. Progressive Access consists in stratifying the information space in different levels of detail through which the user can navigate by means of masking and unmasking mechanisms. We have developed a platform called KIWIS [2] for the design and the generation of adaptable WIS based on a progressive access approach, which implements the five models presented as well as a design methodology. We are currently extending the notion of progressive access towards dynamic adaptability.

REFERENCES

  1. Villanova-Oliver M., Adaptabilit dans les systmes dInformation sur le Web : Modlisation et mise en uvre de laccs progressif, Thse de Doctorat, Institut National Polytechnique de Grenoble, dcembre 2002 (in French).
  2. Villanova-Oliver M., Gensel J., Martin H. & Erb C., Design and Generation of Adaptable Web Information Systems with KIWIS, Third IEEE Conference on Information Technology ITCC-2002 , pp 306-311, Las Vegas, Nevada, USA, April 8-10, 2002.