School of Computer Science and Engineering/Computer Science
Jeffrey S. Rosenschein
Professor

Early Research in Multiagent Systems at Hebrew University

Starting with PhD work at Stanford, and followed by my arrival at Hebrew University, I pioneered a number of approaches to the problems of multiagent systems (MAS) over the years, in conjunction with students and colleagues:

  1. We expanded the focus of Distributed Artificial Intelligence to include interactions among multiple self-interested agents, coining the term benevolent agents to characterize the components of previous DAI systems that had an assumed common interest [references];
  2. We carried out the first research in artificial intelligence (AI) that made use of game theory techniques to analyze multiagent interactions [references];
  3. In separate, follow-on work, we were the first to introduce mechanism design into AI and MAS, exploring issues of incentive compatibility [references];
  4. We were the first to introduce Vickrey-Clarke-Groves mechanisms (and more specifically, the Clarke Tax) into AI, using it to explore multiagent planning [references];
  5. We introduced the Shapley Value into MAS coalition research [references];
  6. We introduced Schelling's notion of Focal Points into AI and MAS, exploring its use in automated interactions. [references]

These techniques have been brought to bear on issues such as agreements reached between automated agents, planning by multiple agents, information revelation in a variety of domains, coalition formation, knowledge representation, and learning. References to this work are given below (a more complete list of my publications can be found here).


The most frequently cited reference is "Rules of Encounter: Designing Conventions for Automated Negotiation Among Computers", Jeffrey S. Rosenschein and Gilad Zlotkin. MIT Press, Cambridge, Massachusetts, 1994. [bibtex-entry] It's based on a series of articles that appeared during the 9 previous years, and includes a lot of the material mentioned below (but the book has better exposition, in my opinion).

Here are the specific references:

"1. We expanded the focus of Distributed Artificial Intelligence to include interactions among multiple self-interested agents, coining the term "benevolent agents" to characterize the components of previous DAI systems that had an assumed common interest;"

"2. We carried out the first research in artificial intelligence (AI) that made use of game theory techniques to analyze multiagent interactions;"

"3. In separate, follow-on work, we were the first to introduce mechanism design into AI and MAS, exploring issues of incentive compatibility;"

  • Negotiation and Task Sharing Among Autonomous Agents in Cooperative Domains, Gilad Zlotkin and Jeffrey S. Rosenschein. The Eleventh International Joint Conference on Artificial Intelligence, Detroit, Michigan, August 1989, pp. 912-917. There are two figures missing from the postscript file; here they are in .bmp format. [bibtex-entry]

  • Negotiation and Conflict Resolution in Non-Cooperative Domains, Gilad Zlotkin and Jeffrey S. Rosenschein. The National Conference on Artificial Intelligence, Boston, Massachusetts, July 1990, pp. 100-105. [bibtex-entry]

  • Incomplete Information and Deception in Multi-Agent Negotiation, Gilad Zlotkin and Jeffrey S. Rosenschein. The Twelfth International Joint Conference on Artificial Intelligence, Sydney, Australia, August 1991, pp. 225-231. [bibtex-entry]

  • A Domain Theory for Task Oriented Negotiation, Gilad Zlotkin and Jeffrey S. Rosenschein. The Thirteenth International Joint Conference on Artificial Intelligence, Chambery, France, August 1993, pages 416-422. [bibtex-entry]

  • The "Rules of Encounter" book describes much of this work. A more "popularized" magazine presentation, based on my IJCAI-93 invited lecture, can be found in "Consenting Agents: Designing Conventions for Automated Negotiation", Jeffrey S. Rosenschein and Gilad Zlotkin. AI Magazine, Volume 15, Number 3, Fall 1994, pages 29-46. Also reprinted in Readings in Agents, edited by Michael N. Huhns and Munindar P. Singh, Morgan Kaufmann Publishers, San Francisco, California, 1997, pages 353-370. [bibtex-entry]

Some of the early work was published in the following journal article:

  • Cooperation and Conflict Resolution via Negotiation Among Autonomous Agents in Noncooperative Domains, Gilad Zlotkin and Jeffrey S. Rosenschein. IEEE Transactions on Systems, Man, and Cybernetics, Special Issue on Distributed Artificial Intelligence. Volume 21, Number 6, November/December 1991, pp. 1317-1324. [bibtex-entry]

There were also journal articles that followed the above conference articles:

"4. We were the first to introduce the Clarke-Groves mechanism (and more specifically, the Clarke Tax) into AI, using it to explore multiagent planning;"

The journal articles that covered this work include:

"5. We introduced the Shapley Value into MAS coalition research;"

"6. We introduced Schelling's notion of Focal Points into AI and MAS, exploring its use in automated interactions"

The journal version of the paper appeared as:


jeff at cs.huji.ac.il
Last modified: 9 February 2004