Recent Changes - Search:

fr en

Research

Teaching

Perso

pmwiki.org
edit SideBar

Phd Work

PhD in Computer Science (September 2006) at the Laboratory LSR (Logiciels, Systèmes, Réseaux) – IMAG (Institut d'Informatique et Mathématiques Appliquées de Grenoble), Université Joseph Fourier (UJF) – Grenoble I, Grenoble, France.

Abstract

The evolution of mobile technologies, like web-enable cellphones, PDAs and wireless networks, makes it now possible to use these technologies for collaborative work through Web-based groupware systems. However, due to the limitations of these technologies and to the mobility that these technologies grant to users, some adaptation is essential in these systems. In order to adapt their behaviour to the user’s context, groupware systems must be built as context-aware systems. Moreover, this adaptation concerns particularly awareness mechanisms since nomadic users have more acute need for adapted awareness information. Dues to the constraints related to new technologies, adaptation ought to act on the contents itself, by selecting it and organizing it according to its relevance for nomadic users. Besides, this relevance depends on the user’s current context, since user’s preferences and needs may change accordingly to this context.

In this work, we present a two-fold approach for adapting awareness information delivered to a nomadic user by web-based groupware systems, through a filtering mechanism which considers both the current context and preferences for this context. We propose an object-based representation for the notion of context, which takes into account the user’s physical context as well as the user’s collaborative context, including elements dedicated to the collaborative process in which the user is involved (notions of group, role, activity, etc.). The user’s preferences are represented by a set of pre-defined profiles, which are exploited by the adaptation process in order to organize the delivered awareness information into several levels of detail, based on a progressive access model. These profiles represent the descriptions of user’s potential contexts and of the filtering rules to apply when the user’s current context matches a potential context. This matching is performed by a set of semantic operations (named equals, contains, and sim, simO and simT similarity measures), which compares both instances of the context model. Then, the proposed filtering mechanism is performed in two steps: first, it analyzes the user’s current context and, thank to these semantic operations, it selects, among the user’s pre-defined profiles, those which have been defined for a situation found in this context. Second, it selects awareness information according to the selected profiles and it uses the progressive access model in order to organize the selected information.

Finally, we propose a framework, named BW-M, which implements our proposition based on an object-based knowledge representation system named AROM, and a Web service, named BWMFacade, that encapsulates this framework for helping groupware designers.
Edit - History - Print - Recent Changes - Search
Page last modified on December 13, 2007, at 11:06 PM