An Anytime MultiAgent System to Manage Electronic Commerce Transactions

Electronic commerce applications are inherently distributed. They must allow actors (sellers and buyers) to trade products thanks to computer systems. These systems must be designed so as to manage distributed information systems and to allow timely negotiation between the actors. That is why decisions must be taken timely in spite of the lack of information. Furthermore, these systems use a lot of information which may be complex and difficult to analyze. So, a promizing approach is to use a multiagent system in which anytime techniques are added. In this paper, we present ANYMAS, a model of real-time multiagent systems applied to the judicious extraction of information from distributed information systems. Then, we give experimental results that show how the system may help sellers and buyers to take good decisions. We particularly show how in anytime techniques, more the time allowed to the computer system is important, more accurate are the results.