Michael N. Huhns & Munindar P. Singh
The agent metaphor comes packaged with a number of powerful abstractions. These include traditional AI abstractions, such as beliefs, knowledge, and intentions as well as the newer and more interesting social, ethical, and legal abstractions.
Traditional AI considers stand alone entities, and studies actions merely in terms of their causes and effects, and simplistic obligations and constraint reasoning over them. In contrast, agents are social, their actions involve the ethical concepts of right and wrong, and they may engage in a variety of legal relations. The latter themes are emphasized in this tutorial.
Agent applications are drawing increasing attention from researchers, practitioners, and even users. All too often, however, current agent-based systems neither support nor use the abstractions that make agents interesting in the first place. Some developers are aware of these limitations, but necessarily come up with ad hoc solutions. However, the applications will benefit from a principled usage of the above abstractions, which enable true agent functionality.
This tutorial includes some historical background, but emphasizes conceptual and technical material relating to the above abstractions. It provides a comprehensive survey, and describes how the above abstractions are being used in some upcoming multiagent systems. This tutorial introduces AI practitioners and students to multiagent systems. It is especially useful for those transitioning from traditional agent systems and applications to multiagent systems proper.
Prerequisite knowledge:
The tutorial is self-contained; it assumes only some familiarity with AI.
The presenters have a long trackrecord in the theory and practice of multiagent systems. They have given a number of tutorials at international computer science conferences. They coedited the book "Readings in Agents", published by Morgan Kaufmann in 1998 and now in its second printing.
Professor Huhns (Ph.D., USC, 1975) edited the
books "Distributed Artificial Intelligence", volumes 1 and 2, and
authored over 100 papers and reports. He is serving or has served on
numerous program committees, conference advisory boards, and journal
editorial boards, including two IEEE magazines and one ACM
transactions.
Homepage: http://www.ece.sc.edu/faculty/Huhns/
Professor Singh (Ph.D., Texas, 1993) authored a
book "Multiagent Systems" and several papers on agents and
databases. He is Editor-In-Chief of IEEE Internet Computing.
Dr. Singh has also chaired conferences on cooperative information
systems and agents.
Homepage: http://www.csc.ncsu.edu/faculty/mpsingh/