The AAAI-2000 Workshop on

Parallel and Distributed Search for Reasoning

held at the National Conference on Artificial Intelligence (AAAI-2000)

July 30-August 3, 2000, Austin, Texas

Many AI systems, such as learners, planners, deduction or expert systems, base their reasoning on intelligently searching in very large search spaces. By intelligently searching, we mean that the system combines various pieces of general knowledge it has with information found so far and uses this combined knowledge to decide where to search next; thus, only limited predictions about the search are possible.

Also, many concepts for systems have been developed to speed up search by using several computing nodes (computers or processors). These concepts must deal with problems like balancing the processor loads or avoiding idle times, bottlenecks, and redundancy. Very often these problems are solved by using a simple search control that is easy to predict.

Thus, it is difficult for systems to use intelligent search while taking advantage of parallel or distributed search. Naturally, this can be fixed by communication between the nodes, but too much communication drastically reduces the gains by the use of several computing nodes. Nevertheless, in the last few years several systems for different application areas have been developed that achieve the combined benefits of using several computing nodes and intelligent knowledge based search. By cooperating, several search systems were often able to solve harder problem instances than each single system could. Although the systems use in most cases domain-specific knowledge, certain types of knowledge and certain parallelization and distribution concepts can be identified that are usable for several application domains.

This workshop aims at bringing together researchers from various application areas that are interested in performing intelligent search using several computing nodes. As a result of this workshop we hope to come to a better understanding of the kinds of applications and knowledge that are well-suited for certain parallelization and distribution concepts and to identify the basic components of such concepts.

Topics

Topics of interest include, but are not limited to:

Workshop format

The workshop will be a mixture of invited talks, submitted presentations and an open discussion with regard to our intended results.

Participation and Submissions

Participation in workshops at the National Conference on Artificial Intelligence is by invitation only and invitees have to register for the main conference. Also, attendance is limited. Therefore, in case that a selection becomes necessary, we ask also researchers that just want to attend the workshop without contributing a paper to send a short email to denzinge@informatik.uni-kl.de expressing their particular interest in the workshop.

Researchers interested in contributing a paper should send it also to denzinge@informatik.uni-kl.de (please send a uuencoded gzipped postscript file). Papers should not exceed 5 pages in the AAAI format. Since we expect that participants of the workshop are interested in rather different application areas, papers should include a brief introduction into the particular application and authors should also try to identify the characteristics that an application area should have, to make it suitable for their approach to parallel and distributed search. We particularly welcome papers that directly address the goals of the workshop: characterizing the kinds of applications and the knowledge that is well-suited for certain parallelization and distribution concepts, and determining the basic components of parallelization and distribution concepts.

The papers will be reviewed by the organization committee (and some additional referees) and all papers of sufficient quality will be included into the workshop notes (and their authors invited to the workshop, of course). Out of these papers several will be selected for presentation at the workshop. The main criteria of this selection will be to cover a broad variety of both application areas and parallelization and distribution concepts and the contribution to the goals stated above. To facilitate a lively and interesting discussion, we will try to make all the papers available to the participants of the workshop before the workshop takes place.

After workshop activities

Depending on the outcome of the workshop and the contributions to it, we plan two activities after the workshop:

  1. Adding to this website (and its mirrors) links to all participants (that agree to this) and their projects related to parallel and distributed search, hopefully, as a result of the workshop, indexed with respect to several criteria. This should provide people who are becoming interested in parallelizing or distributing their intelligent search system with a starting point in the WWW.
  2. Pursuing a formal publication arising from the workshop.

Important dates

Organizing Committee

Jörg Denzinger (Chair)
Computer Science Department
University of Kaiserslautern
Postfach 3049
67653 Kaiserslautern
Germany
FAX: +49-631-205-3558
Phone: +49-631-205-2181
Email: denzinge@informatik.u ni-kl.de 		Yasuhiko Kitamura
Department of Information and Communication Engineering
Faculty of Engineering, Osaka City University
3-3-138 Sugimoto, Sumiyoshi-ku
Osaka 558-8585
JAPAN
FAX : +81-6-6605-3081
Phone: +81-6-6605-3081
Email: kitamura@info.eng.osaka-cu.ac.jp
Eugene Santos Jr.
Department of Computer Science and Engineering
University of Connecticut
Storrs
CT 06269-3155
USA
FAX: 860-486-4817
Phone: 860-486-1458
Email: eugene@engr.uconn.edu		 Bruce Spencer
Faculty of Computer Science
University of New Brunswick
P.O. Box 4400
Fredericton, New Brunswick
CANADA E3B 5A3
FAX: 506-453-3566
Phone: 506-453-4566
Email: bspencer@unb.ca

Last Change: 16/11/99