Emmanuelle Grislin-Le Strugeon

MAPIS, a multi-agent system for information personalization

In the domain of multi-user and agent-oriented information systems, personalized information systems aim to give specific and customized responses to individual user requests. In addition to the ability to analyze user needs and to retrieve, understand and act on distributed data that is offered by any agent-oriented system, multi-agent systems also offer interesting possibilities for interaction, particularly with regard to information sharing and task coordination. Our approach exploits these interactive possibilities in order to make the system capable of personalizing information. In addition, reusable models at both the social and individual levels were chosen for this approach in order to facilitate subsequent information system design. With these two ideas in mind, several models of agent interaction (social) and the internal activity cycles (individual) have been proposed with the aim of creating a multi-agent system for information personalization.

Flexible Hierarchical Organisation of Role Based Agents

MAS are particularly adapted to deal with distributed and dynamic environment. The management of business workflow, or data flow, is typically a good application field for them. This kind of application requires centralization of the data control and flexibility to change of the network. By taking inspiration from the Holonic Multi-Agent organisations (that are particular pyramidal organisations where agents of a layer (having the same coordinator) are able to communicate and to negotiate directly between them), and working mechanisms of the complex human organisations, we propose a framework for holonic multi-agent organizations, able to growth and distribute roles and where agents have adaptable roles.

A methodology to bring MAS to information systems

Agents have been used to provide qualities of adaptation to the Information Systems (IS) since the Nineties. Apart from agent-based IS, existing IS can benefit from additional services provided by agents. We focus on personalization services, which aim is to give specific and customized responses to individual user requests. This paper describes a methodology to design the adaptation part of an IS with the help of a multi-agent system (MAS). An initial multi-agent architecture with an application-independent organization and interaction protocols is provided. On this base, the methodology proposes a guide into the design of the specific knowledge and skills that allow adaptation of the agents behaviour to the characteristics of the application, and into realization of the integration between IS and MAS. An example taken in the transport information field illustrates our proposition.

Behavioral self-control of agent-based virtual pedestrians

Agents need some kind of autonomy to act in open and dynamic environments. However, the resulting behavior is not always well adapted to the situation. The use of models that combine different reasoning levels creates other difficulties. The management of the control transfer and balance between abstract time-consuming reasoning processes and reactive quick low-level behaviors are particularly complex. Our model is based on the self-observation and control of agents. The main topic is the inconsistency detection used to transfer the control between contiguous levels. This is detailed in the context of virtual pedestrians design.

Personalised Information Systems

Abstract This article presents a multi-agents architecture to build personalised information systems. This architecture is composed of four parts: the first one interacts with the user, the second one coordinates the process of the response, the third one manages the user model and the last one searches the data.

An Energy-Aware Learning Agent for Power Management in Mobile Devices

The optimization of the energy consumption in mobile devices can be performed on hardware and software components. For example, reducing the screen brightness or switching off GPS. The energy control must take into account both the current context and user habits, on the base of usage knowledge acquired from sensors and OS data records. The whole process of energy management then includes data collection, usage learning and analysis, decision-making and control of device components. To integrate these activities, we propose to use a software agent whose goal is to save the energy of the mobile device with the lowest effect on QoS.

MAS architecture and knowledge model for vehicles data communication

Completely autonomous vehicles in traffic should allow to decrease the number of road accident victims greatly, and should allow gains in terms of performance and economy. Modelling the vehicles interaction, and especially knowledge sharing, is one of the main challenges to optimize traffic flow with autonomous vehicles. We propose in this paper a model of knowledge communication between mobile agents on a traffic network. The model of knowledge and of interaction enables to propagate new knowledge without overloading the system with a too large number of communications. For that, only the new knowledge is communicated, and two agents communicate the same knowledge only once. Moreover, in order to allow agents to update their knowledge (perceived or created), a notion of degradation is used. A simulator has been built to evaluate the proposal, before to implement it in mobile robots. Some results of the simulator are proposed in this article.