DAI-List Digest Monday, 7 December 1992 Issue Number 99 Topics: New Book on DAI Theoretical DAI??? Re: Theoretical DAI??? Administrivia: DAI-List now conforms to the RFC1153 Digest Format, so that it can be expanded by M-x undigestify in emacs and zmail. Please send submissions to DAI-List@mcc.com. Send other requests, such as changes in your e-mail address, to DAI-List-Request@mcc.com. ---------------------------------------------------------------------- Date: Wed, 25 Nov 1992 12:24:05 +0100 From: Frank von Martial Subject: New Book on DAI Book DAI Book You may wish to note the following book on DAI: COORDINATING PLANS of AUTONOMOUS AGENTS by Frank von Martial DETECON GmbH Germany appeared as "Lecture Notes in Artificial Intelligence", No. 610 Springer Verlag, June 1992 ISBN 3-540-55615-x 250 pages A novel approach to coordinate activities of autonomous agents is presented. The coordination of plans is triggered by the relations which exist between the actions of different plans. The main contributions of this book include: 1. a detailed taxonomy of relations between the plans of different agents, 2. strategies for coordination and plan modifications based on these relations, and 3. a negotiation framework for synchronizing activities. The reader may also appreciate the thorough introduction to the field of distributed AI and its subarea, multiagent planning (ca. 50 pages). CONTENTS: Foreword (by Victor R. Lesser) 1. Introduction 2. Cooperation and Coordination of Intelligent Agents 3. Actions and Plans in Multiagent Domains 4. Multiagent Plan Relations 5. Handling Plan Relations 6. Negotiation for Plan Coordination 7. A Synthesized Model of Distributed Planning by Autonomous Cooperating Agents 8. Plan Coordination for Office Automation and Computer-Supported Cooperative Work 9. Conclusions ------------------------------ From: Mack D P Date: Thu, 26 Nov 1992 07:04 CST Subject: Theoretical DAI??? This is an attempt to start a debate on the nature and role of theory within DAI. I would like to see if we can come to any kind of consensus on the following: 1) Is theory useful within DAI? 2) What kinds of things do we expect our theories to talk about? 3) Does DAI really have a common theoretical core? 4) What methodology should we adopt to improve our theories? My own view is that at present theory is of limited use in DAI, mainly because it is not sufficiently developed. The type of use I see being made of such theories is as a repository for our understanding of the common core of matters of interest to the DAI community. I believe that such a common core will be found in a fairly coarse-grained model of agenthood and interaction, but one where agents have some relevant internal structure. The last question is one I cannot easily answer. I accept that this seems a very waffly start - but as far as I know it is this type of general discussion for which this group was created. I would appreciate responses from anyone who disagrees with any of the statements I have made, or wants to expand upon any of the points. ------------------------------ From: Mack D P Date: Fri, 27 Nov 1992 09:43 CST Subject: Re: Theoretical DAI??? It has been pointed out to me in an e-mail reply that there are of course many disciplines such as economics, sociology and distributed systems which share (or appear to share) some of the same concerns with DAI. My point here is that the *types* of theories developed in these disciplines are completely different. Sociology tends to rely on theories expressed in natural language; economics on real variables; distributed systems on process calculi. Which type of theory should we be going for in DAI and why? Which of these theories generates the most insights? Which is of most practical value? ------------------------------ End of DAI-List Digest Issue #99 ********************************