A software agent community model for the shopping assistant agent system

In recent years, the ubiquitous computing model of Personal Digital Assistants (PDAs) has attracted many researchers' attention. However, PDAs also have limitations, such as slower computing power, smaller screens, and smaller storage capacity. This thesis examines the advantages of using Mobile Agent (MA) technology in PDA application development to attempt to compensate these limitations.;A Shopping Assistant Agent System (SAAS) was implemented as an experimental model of the proposed system structure. The SAAS aims to provide an assistant for people to do their routine shopping in supermarkets by using mobile agent technology in a PDA. During the server-side SAAS design and implementation, a number of problems that challenge the success of the SAAS were encountered due to the distinctive characteristics of agent-based services. To solve these problems, the concept and architecture of Software Agent Community (SAC) were proposed to manage distributed agents, agent platforms and MAS applications.;To demonstrate the advantages of this hybrid system structure and the concept of SAC, the TEEMA (TRLabs Execution Environment for Mobile Agent) Agent Community and the SAAS server-side system were implemented and deployed.;As a relative new technology, agent technology has not been widely adopted in industrial applications due to the integration cost between legacy software and agent applications. This thesis proposes a hybrid system structure, which is a combination of MA technology and Java 2 Platform Enterprise Edition (J2EE) technology, to provide a cost-efficient approach to use agent technology in enterprise-level applications.