Developing knowledge models for multi agent system is currently ad-hoc. Neither guidelines nor mechanisms nor techniques exist for developing the agent knowledge. This paper presents a conceptual architecture and implementation for a knowledge model for a class of multi-agent mediator systems. They arise from the prototyping of a multi agent system finding an adviser for an international student at an Australian University. The knowledge model development process is based on the quality criteria for knowledge based system of applicability and maintainability, as well as issues required by agents in dealing with knowledge. The process components are described and form the requirements for an agent developer in dealing with the development of agent knowledge from scratch.
Lecture notes in computer science: agent computing and multi-agent systems: proceedings of the 10th Pacific Rim International Conference on Multi-Agents, Bangkok, Thailand, 21-23 November 2007 / R. Goebel, J. Siekmann, and W. Wahlster (eds.),
Vol. 5044, pp. 440-445