DAI-List Digest Thursday, 6 June 1991 Issue Number 37 Topics: Report available on "Dynamics of Expectations" Paper available on "Teamwork" Report available on "Problem Solving by Committee" CFP: Knowledge and Action at Social and Organizational Levels CFP: IJPRAI Special Issue on Blackboard Systems 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: Tue, 28 May 1991 16:28:29 PDT From: Bernardo Huberman Subject: Report available The following preprint/ report is available. "Dynamics of Expectations" N. S. Glance and B. A. Huberman Dynamics of Computation Group Xerox PARC Palo Alto, CA. 94304 Abstract: The dynamical behavior of interacting agents that make decisions based on expectations about the future, along with imperfect knowledge about the present and the past, is studied within a model that incorporates transaction costs and discounted payoffs, essential features of any economic organization. Long forecasts tend to destabilize the system, and a phase diagram is obtained that exhibits a sharp boundary separating stable equilibria from complex dynamics as the horizon length increases. The existence of an optimal strategy, in the sense that profits are maximized and behavior is stable, is also demonstrated. Rising expectations in time are shown to lead to a crash whereby the fixed point gives way to wild oscillations. This behavior is typical of two types of expectations: flat ones, and those that follow short term trends. It is also shown how a diversity of expectations among the agents, coupled to reward mechanimsms, can generate overall dynamics characterized by cyles of almost stable behavior interrupted by sudden crashes. ------------------------------------------------------------------------ Date: Tue 28 May 91 17:09:28-PDT From: Phil Cohen Subject: Paper available Teamwork Technote 504, March 1991 Phil Cohen Artificial Intelligence Center SRI International and Hector Levesque Dept. of Computer Science University of Toronto ABSTRACT What is involved when a group of agents decide to do something together? Joint action by a team appears to involve more than just the union of simultaneous individual actions, even when those actions are coordinated. We would not say that there is any teamwork involved in ordinary automobile traffic, even though the drivers act simultaneously and are coordinated (one hopes) by the traffic signs and rules of the road. But when a group of drivers decide to do something together, such as driving somewhere as a convoy, it appears that the group acts more like a single agent with beliefs, goals, and intentions of its own, over and above the individual ones. But given that actions are performed by individuals, and that it is individuals who ultimately have the beliefs and goals that engender action, what motivates agents to form teams and act together? In some cases, the answer is obviously the inherent value in doing something together, in part because the actions simply cannot be performed alone. But in many cases, team activity is only one way among many of achieving the goals of the individuals. What benefits do agents expect to derive from their participation in a group effort? In this paper, we attempt to provide answers to these questions. In particular, we argue that a joint activity is one that is performed by individuals sharing a certain specific mental property, namely that of a "joint intention." We show how joint intentions affect and are affected by beliefs, goals, commitments and intentions of the participants. A significant complication arises when one considers that joint intentions should be subject to similar demands for stability over time as are individual intentions. Our attempts to meet this demand are presented. Regarding the benefits of teamwork, we show that in return for the overhead involved in participating in a joint activity, an agent expects to be able to share the load in achieving a goal in a way that is robust against certain possible failures and misunderstandings. To obtain this paper, please send an email request to Wendy Hall (wendy@ai.sri.com). ------------------------------------------------------------------------ Date: Wed, 5 Jun 1991 15:00:47 PDT From: Bernardo Huberman Subject: Report available "Problem Solving by Committee" S. H. Clearwater, B. A. Huberman and T. Hogg. Dynamics of Computation Group Xerox PARC Palo Alto, CA. 94304 Abstract: It is widely believed that a group of cooperating agents engaged in problem solving can solve a task faster than either a single agent or the same group of agents working in isolation from each other. Nevertheless, little is known about the quantitative improvements that result from cooperation. We present a number of experimental results on constraint satisfaction that both test the predictions of a theory of cooperative problem solving and provide a quantitative assessment of the value of cooperation. These experiments clearly exhibit a universal improvement in performance that results from cooperation, and suggest an alternative methodology to existing techniques for solving constraint satisfaction problems in computer science and distributed artificial intelligence. ------------------------------------------------------------------------ Date: Thu, 30 May 91 15:16:04 +0200 From: gasser@zeta.ibp.fr (Les GASSER) Subject: Reminder - CFP CALL FOR PARTICIPATION KNOWLEDGE AND ACTION AT SOCIAL AND ORGANIZATIONAL LEVELS AAAI Fall Symposium November 14-17, 1991 Asilomar, CA Research on representing and reasoning about knowledge and action at analytical levels more aggregated than that of the individual---topics such as social and organizational foundations of knowledge, multiple-perspective reasoning, consensus, multiple rationality, coordination and collaboration, commitment, stable organization, representing groups and group activity, etc.---is becoming increasingly important. Such research influences many fields, including cognitive science, distributed systems, design, human-computer interaction, natural language processing, computer-supported cooperative work, basic AI and distributed AI, social studies of science, and the engineering of AI systems. This symposium will address a number of relevant foundational scientific issues, such as: CONCEPTUAL MODELING: How can agents, knowledge, commitment, etc. be modeled at the social, the individual, and at integrated levels? Can a group be a locus of knowing, rather than an individual? How can groups model their own activities and influence other groups, even in so-called "open systems?" INTEGRATING MULTIPLE PERSPECTIVES: How do actors reason among multiple representations? How can actors deal with conflict in knowledge and action? How can actors dynamically construct useful meanings for the objects and actions that are part of their joint activity? INTEGRATING SITUATED AND GENERAL KNOWLEDGE: How can agents and groups address the basic tension between local, "situated" knowledge and action, and the non-local conception of general knowledge and action-at-a-distance? How can an actor influence another actor, e.g. by sending a message, without global semantics or control? INTEGRATING SOCIAL AND INDIVIDUAL LEVELS OF ANALYSIS: What would a "middle ground" theory, that integrates both individual agency and social structure, be like? (How) can we make middle-ground theories computational? COMMON GROUNDS FOR INTERACTION: What are the foundations for mutually compatible languages, assumptions, and other common bases for interaction? How is global or shared knowledge possible? How can agents "stay out of each others' way?" METHOD: What are appropriate research methods for addressing these questions in principled ways? What are the roles and limitation of data from biological, human, and social studies? What are the roles and limitations of current formal models and theories? Prospective participants should contact a symposium committee member for a more detailed description of goals and issues. Then they should submit a 3 to 4 page description of their work, and some indication of their relevant background. Ideally, presentations would integrate theoretical perspectives and practical experience, addressing either the scientific issues in some space of empirical phenomena, or engineering concerns in applications domains. Submissions should be sent BY ELECTRONIC MAIL to hovy@isi.edu to arrive by June 14, 1991 and will be promptly acknowledged. If electronic mail is impossible, send four paper copies, clearly marked "AAAI FALL SYMPOSIUM" to arrive by June 14 to: Dr. Eduard Hovy USC / ISI, 4676 Admiralty Way, Marina del Rey, CA 90292-6695 Fax: (213) 823-6714. Program Committee: Phil Agre (University of Sussex) [agre@cogs.sussex.ac.uk] Danny Bobrow (Xerox PARC) [bobrow@parc.xerox.com] Les Gasser (USC) (Chair) [gasser@usc.edu] Jim Hendler (University of Maryland) [hendler@cs.umd.edu] Eduard Hovy (ISI) [hovy@isi.edu] Ed Hutchins (U. C. San Diego) [ehutchins@ucsd.edu] Leigh Star (University of Keele) [soa03@keele.ac.uk] ------------------------------------------------------------------------ Subject: IJPRAI CFP Date: Mon Jun 3 10:44:36 1991 From: netearth!mksaxena@shakti.ernet.in Call for Papers and Referees for a Special Issue of International Journal of Pattern Recognition and Artificial Intelligence on Blackboard Systems (BBS) The blackboard architecture has emerged as a dominant problem solving paradigm. Keeping this in view International Journal of Pattern Recognition and Artificial Intelligence (IJPRAI) has planned to devote the September 1992 issue to Blackboard Systems. Prospective authors are invited to submit tutorials, survey, critiques, case study, applications, or pedagogic manuscripts. Topics of interests include, but are not limited to : - Organisation and Control of BBS - Development Environments and tools for BBS - Architectures to support Concurrency and Parallelism - Languages and Support Tools for Distributed Blackboard Systems - Experimental Systems and Case Studies - Real-time Systems, Transaction oriented systems, decision support systems - Applications in Pattern Recognition, Image Processing, Computer Vision, Robotics. Instructions for Submitting Manuscripts Authors are requested to submit six copies (in English) of about 6,000 words (approximately 25 double-spaced pages using a 12-point type) including all text,figures, and references. Manuscripts should have first page as title page containing : paper title, full name, affiliation, postal address, e-mail address, telephone, and fax number of all authors, a 500 word abstact and a list of 5-10 keywords. Manuscripts must not have been previously published or currently under consideration for publication elsewhere. Send Submissions and Queries to Prof. P.C.P. Bhatt Manoj K. Saxena Dept. of Computer Science & CMC Ltd. Engineering A-5, Ring Road Indian Institute of Technology N.D.S.E. - I Hauz Khas New Delhi - 110049, India. New Delhi - 110016, India. Tel : 091 (11) 686-5873 Tel: 091 (11) 462-3111 Fax : 091 (11) 686-8765 Fax: 091 (11) 684-4652 Email : pcp@netearth.ernet.in Email: mksaxena@netearth.ernet.in Deadlines - 500 word Abstract of the Manuscript November 15, 1991 - Six (6) Copies of the Full Manuscript January 1, 1992 - Notification of Decisions April 15, 1992 - Final Version of the Manuscript June 1, 1992 - Date of Special Issue September 1992 Referees If you are willing to referee papers for the special issue, please send a note with research interest to either of the guest editors or to Editors-in-charge Prof. P.S.P. Wang/ Prof. H. Bunke at the following address Prof. P.S.P. Wang Prof. H. Bunke Editor-in-Charge, IJPRAI Editor-in-Charge, IJPRAI Northeastern University Institut for Informatik College of Computer Science und Angewandte Mathematik 360 Huntington Avenue Universitat Bern Boston, MA 02115, USA Langgassstrasse 51 CH-3012, Bern, Switzerland Tel : (617) 437-3711 Tel : (031) 658681 Fax : (617) 437-5121 Fax : (031) 653965 Email : pwang@corwin.ccs.northeastern.edu Email : bunke@iam.unibe.ch