Bringing Multi-agent Systems into Human Organizations: Application to a Multi-agent Information System

Abstract

Agents are increasingly used to search for information on the Internet. However, they are principally used as single agents and not as part of a multi-agent system. Indeed, few projects use agents that communicate or collaborate with each other. This lack of communication often causes users to be isolated in front of their computer. We think that it is necessary for the user of an information search system (like an actor of a technological watch cell or a researcher in a laboratory for example) to be aware of what his/her colleagues are searching for (or at least to have knowledge of a part of their searches). This should avoid redundancies of information and work, and should (re) generate a team feeling among the actors. That is why we propose, in this article, an multi-agent information system, which is composed itself of multi-agent systems located on the computer of each actor of a technological watch department, and which collaborate with each other and with the actors. This multi-agent architecture has been chosen in agreement with actors in a real department, following the analysis and modeling of their activities. The method of integrating a multi-agent system into a human organization is also discussed in this article.