Wided Chaari

ABSTRACT This paper deals with the evaluation of multiagent systems. The related works in this issue are most often interested in evaluation of agent design methodologies or agent tools and platforms but do not address sufficiently evaluation of multiagent applications from a general point of view, the used evaluation criteria remain either specific to a special topic or application dependent. In this work we present an evaluation approach to measure general characteristics of multiagent systems. The proposed method follows a three step process: observation, modelling and measure. The modelling technique is based on graph theory which is used to define and estimate metrics for the evaluation of structural properties of multiagent systems. In this paper, we focus especially in organization as evaluation criterion as it is a very interesting characteristic transposing the social dimension of multiagent systems. Therefore, specific metrics based on graph theory are proposed to analyse agent organizational structures. The tests and experimentations are carried out on a multiagent application for the simulation of production planning and control in supply chains.

Program Supervision aims at automating the use of complex programs, independently of any particular application domain. Program Supervision Knowledge-Based Systems (KBS) offer original techniques to plan and control program processing activities. The distribution of such systems becomes essential because real applications imply more and more participants on various sites. It allows either to simply consult existing distant knowledge bases on the use of programs, or to collaboratively construct new knowledge bases, or to launch a request on a distant KBS with local data. Our current application concerns assistance to physicians in the use of medical imagery programs and more precisely osteoporosis detection in bone radiographies. The image processing request management based on several distant programs is transparent to physicians. In this paper, we propose a distributed architecture based on mobile agents for program supervision systems and we show why and how to use mobile agents f...

The objective of our work is to study various distribution methods for knowledge-based systems (KBS) allowing access to (several) remote knowledge bases as well as remote auxiliary programs or inference engines. In particular, we are interested in knowledge-based systems for a task named program supervision, with an application to medical imagery. Program supervision KBSs allow to automation the use of complex programs, independently of any particular application domain. They offer original techniques to plan and control program processing activities. The distribution of such systems becomes essential because real applications imply more and more participants on various sites. Given distributed data, programs and knowledge, our aim is to propose convenient and efficient models of distributed program supervision systems. Our current application relates to medical assistance in the use of medical imagery programs for osteoporosis detection in bone radiographies. A distributed approach...

ABSTRACT Industrial systems are complex, distributed and need high-quality tools for their control and monitoring. Although, multiagent systems (MAS) are well adapted to such applications, there is still a gap between academic research on MAS and their potential use on a large scale as in industry. Such domain, with important economic stakes, requires high-performance and reliable applications. However, performance issues were rarely discussed in multiagent environments; the existing works focus only on evaluation of multiagent design methodologies and development platforms. In this paper, we identify, among the MAS features, those that make them suitable to industrial applications. Then, we give an overview on MAS evaluation and we propose a measurement model based on the graph theory to analyse several communication facets in MAS. An experimentation is carried out on two architectures of a multiagent diagnosis application. The comparison of the results points out the correlation between the measurement model and the communication network.

Although the multiagent technology is being more and more used in various application domains, it is still suffering from lacks related especially to performance issues. In order to analyse a MultiAgent System’s performance, we need to observe its behaviour so that to be able to measure its main behavioural characteristics such as communication, autonomy, adaptation, etc. Thus observation in MAS should clearly address the nature of the needed information and the manner it should be observed. In this paper, a general approach to observe multiagent applications is proposed for a performance evaluation purpose. The solution is based on Aspect Oriented Programming which is a new technology of Software Engineering allowing a better modularization and reutilisation of programs. Observation is performed through an aspect-based layer which detects and traces significant events in the observed MAS’s execution.

Résumé Nous nous intéressons aux Systèmes à Base de Connaissances (SBC) en pilotage automatique de programmes dont le but est d'automatiser l'utilisation de logiciels complexes, grâce à des systèmes capables de planifier des traitements et d'en contrôler l'exécution. Leur distribution devient indispensable car les applications en grandeur réelle impliquent de plus en plus de participants inter ou intra-entreprises sur divers sites. La distribution permet soit la simple consultation distante de bases de connaissances, soit la construction collaborative de bases de connaissances, soit le lancement d'une requête sur un SBC distant, avec des données locales. L'application actuelle porte sur l'assistance de différents médecins en Imagerie médicale et plus précisément pour la détection de l'ostéoporose dans des radiographies d'os de leurs patients. L'assistance vient du fait que le traitement d'une requête sur des images données, par plusieurs p...

Many MultiAgent Systems (MAS) have been developed in various application domains such as computer networks, Internet, industrial applications, automation, process control, air traffic, robotic, simulation, etc. In spite of the rapid growth of the international interest in MAS field and the importance of the number of developed MAS, there is still a lack related to their performance evaluation. In fact,

Program supervision aims at automating the use of complex programs, independently of any particular application domain. Program supervision systems offer original techniques to plan and control program processing activities. The distribution of such systems becomes essential because real applications imply more and more participants on various sites. Therefore, given distributed data, programs and knowledge, our aim is to assist distant users, who are not necessarily specialists in image processing, in the latter task (image processing) and then in their decision making according to some parameter values and to their experience while observing the resulting images. Our current application relates to medical assistance in the use of medical imaging programs for osteoporosis detection in bone radiographies. A distributed approach will allow executing distant physician requests using an existing system and will favor collaborative work and sharing of knowledge to build new knowledge-based systems. The objective of this paper is to present a distributed system of computer-aided diagnosis and to show how a distributed program supervision system can assist distant physicians in their diagnosis so that the management of their requests remains transparent to them.

Abstract: The current global dimension of human exchanges in any domain (work, commerce, learning, entertainment…) is accompanied by technologies that enhance synchronous and asynchronous communication thus facilitating both collaboration and competition: the two driving forces for progress since ages.

Collaboration is the basis for conceiving, coordinating and implementing the tasks associated to complex goals. Collaboration is pervasive: there is practically no human challenging domain that is not influenced by collaborative processes, in particular Education in formal and informal settings. For these reasons we have to consider collaboration at a distance as a “new” key phenomenon that deserves to be studied, thus modeled in order as much as possible to foresee its effects.