Evolving software agents in e-commerce with GP operators and knowledge exchange

Abstract

Agent-based electronic commerce (e-commerce) has attracted more and more efforts in academic and industrial fields. However, little work has been devoted to exploring the learning and evolving capability of software agents. This paper addresses issues of evolving software agents in e-commerce. Agent structure and organization are introduced to facilitate the evolution process, and agent fitness and life cycle are employed as evolution mechanisms. Two approaches are proposed to equip agents with evolution features. One approach is to use Genetic Programming (GP) operators to restructure agents. Another one is to enhance the knowledge of agents by exchanging with other agents.