CALL FOR PAPERS / PARTICIPATION
The biannual Workshop on Nonmonotonic Reasoning, Action, and Change (NRAC) was first established in 1995. Since its inception it has been held in conjunction with the International Joint Conference on Artificial Intelligence (IJCAI).
An intelligent agent exploring a rich, dynamic world needs cognitive capabilities in addition to basic functionalities for perception and reaction. The abilities to reason nonmonotonically, to reason about actions, and to change one's beliefs, have been identified as fundamental high-level cognitive functions necessary for common sense. Research in all three areas has made significant progress during the last two decades of the past century. It is, however, crucial to bear in mind the common goal of designing intelligent agents. Researchers should be aware of advances in all three fields since often advances in one field can be translated into advances in another. Many deep relationships have already been established. This workshop has the specific aim of promoting cross-fertilization. The interaction fostered by the biannual NRAC has helped to facilitate solutions to the frame problem, ramification problem, and other crucial issues on the research agenda.
Much recent research into reasoning about actions has been devoted to the design and implementation of languages and systems for Cognitive Robotics. Successful case studies demonstrate the applicability of these results for furnishing autonomous robots with high-level cognitive capabilities that enable plan-oriented behavior. Advancing the field of Cognitive Robotics, current research in reasoning about actions focuses on two crucial aspects of robots acting in open, real-world environments: Reasoning about knowledge and belief, and dealing with a challenge known as the qualification problem.
Autonomous, mobile robots choose most of their actions conditioned on the state of their environment. As their information about the world state is generally limited, robots are equipped with sensors for the purpose of acquiring information about the external world. The use of sensing actions is often an integral part of a successful plan, and in order to devise these plans robots need an explicit representation of what they believe the world looks like and how sensing affects their beliefs. Moreover, the execution of a plan needs to be constantly monitored and beliefs have to be revised in accordance with new observations. One goal of the workshop is to bring together researchers from the two areas of reasoning about actions and theory change, in order to join their effort of developing theories and designing systems for intelligent use of sensors and belief revision.
Intelligent agents acting in open environments inevitably face the qualification problem, that is, the executability of an action can never be predicted with absolute certainty; unexpected circumstances, albeit unlikely, may at any time prevent the agent from performing an intended action. Planning and acting under this proviso requires the agent to rigorously assume away, by default, all of the numerous possible but unlikely qualifications of their actions, lest the agent be unable to devise plans which, although not guaranteed of success, are perfectly reasonable. Assuming away unlikely but not impossible qualifications means that, if to the surprise of the agent an action actually fails, then the default conclusion should no longer be adhered to. In this respect the entire process is intrinsically nonmonotonic, which shows the increasing importance of pursuing the interrelation between reasoning about actions and nonmonotonic reasoning.
Comparing and contrasting our current formalisms for nonmonotonic reasoning, reasoning about action and belief revision helps identify the strengths and weaknesses of the various methods available. It is an important activity that allows researchers to evaluate the state-of-the-art. Indeed a significant advantage of using logical formalisms as representation schemes is that they facilitate the evaluation process. Moreover, following the initial success, more complex real-world applications are now within grasp. An implementational testbed is a primary means by which existing theories of nonmonotonic reasoning, action, and change are evaluated. Experimentation with prototype implementations not only helps to identify obstacles that arise in transforming theoretical solutions into operational solutions, but also highlights the need for the improvement of existing formal integrative frameworks for intelligent agents at the ontological level.
For the purpose of developing practical solutions to real-world problems, some obvious questions arise: What nonmonotonic logics and what theories of action and change have been implemented? Which ones are implementable? What can be learned from existing applications? What is needed to improve their scope and performance?
Despite the progress over the last few years, there remain the above mentioned and other difficult open problems for theories of nonmonotonic reasoning, action, and change. We hope to explore new approaches to these problems during this workshop.
This workshop will bring together researchers from all three areas with the aim to:
There will be a special theme at this year's NRAC: We especially encourage submissions describing solutions to the Wumpus World--a problem for logically reasoning agents--based on existing systems for reasoning about actions. Details can be found on the Logically Reasoning Agents - Problem Page.
We plan to devote a session of the workshop on papers presenting solutions to this problem and accompany the presentation of these papers by a panel on Comparative Evaluations of systems for logic-based agents.
The Programme Committee invites submissions of long papers describing new results of length no more than 12 pages of standard LaTeX 12pt article format. The format of the papers may be in either hardcopy form (send 4 copies) or electronic form (postscript only). The Programme Committee has a preference for electronic submissions. To submit a paper electronically, the authors should send an email with subject "NRAC'05 Submission" to both co-chairs (see below), with the file of the paper as an attachment (preferably Postscript format), and the following information in the body of the email in plain text:
The Programme Committee also welcomes suggestions for panels. These may be submitted by email to the Workshop co-chairs.
Deadline for Submission: April 11, 2005 Notification of Acceptance: May 9, 2005 Compilation of Participant List: April 15, 2005 Final version due: May 30, 2005 Workshop: August 1, 2005
Postscript files should be sent to
firstname.lastname@example.org and email@example.comHardcopy papers should be sent to one of the Workshop Chairs. All submissions should include email address, and postal address of all authors. Deadline dates will be strictly adhered to.
Leora Morgenstern (Co-chair)
IBM T.J. Watson Research Center
30 Saw Mill River Road, Mail Stop H1B54
Hawthorne, NY 10532, USA
Maurice Pagnucco (Co-Chair)
School of Computer Science and Engineering
The University of New South Wales
Sydney NSW 2052 AUSTRALIA
Tel: +61-2-9385 6925
Fax: +61-2-9385 5995
Gerhard Brewka University of Leipzig (Germany) Norman Foo University of New South Wales (Australia) Michael Gelfond Texas Tech University (USA) Gerhard Lakemeyer Aachen University of Technology (Germany) Jerome Lang Universite Paul Sabatier (France) Vladimir Lifschitz University of Texas at Austin (USA) Fangzhen Lin Hong Kong Univ. of Science & Tech (China) Leora Morgenstern IBM T.J.Watson Research Center (USA) Maurice Pagnucco University of New South Wales (Australia) Pavlos Peppas Patras University (Greece) Erik Sandewall Linkoping University (Sweden) Michael Thielscher Dresden University of Technology (Germany) Mary-Anne Williams University of Technology, Sydney (Australia)
Attendance at the Workshop will be limited and by invitation only. Authors of accepted papers will be invited. Others wishing to attend will be selected on the basis of their track record, and should submit a brief description of their work and interest in this area (including a short list of publications). Participants must register for the main IJCAI conference. IJCAI will charge an additional fee for workshop particiption.