Application of Learning Mechanism in Agent-Based Automatic Negotiation Technology

Abstract

Business negotiation is a key technique for the development of electronic commerce. In order to improve trade efficiency and reduce trade cost, it is necessary to realize automatic negotiation or half-automatic negotiation during the electronic trade, and the agent technique may achieve these functions. In the multi-agent systems (MAS), the theory of learning games is one of important methods for agent negotiation. During negotiation between Buyer side and seller side, the trade which is whether successful or not depends on their trade strategy model to a great extent. Therefore, in this paper, we use the learning game to discuss the negotiation of e-business, and an agent-based negotiation model is presented by using a new learning algorithm. We specify the agents how to do learning efficiency based on the change of opponents’ strategies and negotiation setting, then readjust their optimal strategy in order to maximize their payoff. In last, a correct trade strategies model is proven through an actual example.