A multi-agent system for electronic commerce including adaptive strategic behaviours

Abstract

This work is primarily based on the use of software agents forautomated negotiation. We present in this paper a test-bed for agents in anelectronic marketplace, through which we simulated different scenariosallowing us to evaluate different agents negotiation behaviours. The systemfollows a multi-party and multi-issue negotiation approach. We tested thesystem by comparing the performance of agents that use multiple tactics withones that include learning capabilities based on a specific kind ofReinforcement Learning technique. First experiments showed that the adaptiveagents tend to win deals over their competitors as their experience increases.