Revised on Nov. 17, 2001
In the next decade some of the global economy is expected to run on the Internet, where some information resources will be connected among/inside companies and people. Instead of human, software agents, which are autonomous and intelligent problem solvers that reason deliberately using rich knowledge, will be deputized in some areas of business, management/control, education, etc. by exchanging information services and goods with other agents , which are emerging as "e-commerce." In the future, we envision that agents will form communities or organizations for automated business in order to achieve mutual coordination or agreement with other agents, as delegates of companies or persons.
Multiagent system technology is therefore crucial for this type of application. Each agent has a shared goal or a goal that may be different from but interfere with other goals. In either case, the task of an agent often consists of a number of subtasks, some of which have to be done by other agents. Assigning these subtasks to appropriate agents while avoiding resource conflicts is a central issue of coordination. The coordinated activities are determined partly based on the task relations (e.g., for avoiding conflicts or understanding task dependencies) and partly based on the agent's organization (e.g., for understanding which agent a task is assigned to). An organization can also facilitate the agents' efficient collaborations by keeping reliable partners for quality level, low cost, and laborsaving without overhead .
Organizational decisions are, however, sophisticated. For example, an agent that is a delegate for a person or a company must take into account relations that reflect actual real-world relations when assigning tasks; and agents' characteristics, such as reliability and timeliness. In a dynamic environment, furthermore, organizational information changes over time while an agent is inactive. This is because an agent starts/stops operations or is discarded; it autonomously enters or leaves a group; it is restarted with new knowledge; and it must adapt to the real-world relations. Besides the organization must constantly be refreshed for more effectiveness by introducing new and better agents and by ejecting relatively costly or less reliable agents. Thus, maintaining organizational information for coordination is a challenging issue.
However cooperative reasoners, planners, and schedulers with a mechanism for maintaining organizational information can easily become complicated. To tackle these issues, we think that knowledge about the domain and (non-local) task relations  should be separated from organizational information. Our system, which is called the organizational information maintenance and assisting system (OIMAS), works like a secretary of an agent, maintains updated organizational information that accurately reflects real-world/Internet objects and gives the agent suggestions for ensuring good coordination.
This research does not focus on any particular coordination model nor how this information is used. OIMAS is intended to be used in applications so that complex decisions can be made simpler. It maintains some of the macro- and meso-level organizational information ; micro-level information, such as task dependencies for coordinated activities, is beyond the scope of this research. Because organizational information is often changeable, our purpose is to retain and use the cyberspace organization consisting of multiple agents for responsible real-world applications.
Topics in this research are:
- how OIMAS and the problem solver in an agent are designed so that the mechanism for generating coordination activities can be shared,
- how the organizational information is represented, and
- how this framework can be applied to actual applications.
 V. Lesser, B. Horling, A. Raja, X. Zhang and T. Wagner. "Resource-Bounded Searches in an Information Marketplace," IEEE Internet Computing, Vol. 4, No. 2, pp. 49 - 58, 2000
 S. Willmott and B. Faltings. "The Benefits of Environment Adaptive Organisations for Agent Coordination and Network Routing Problems," Proc. of ICMAS2000.
 K. Decker and V. Lesser: "Quantitative Modeling of Complex Environments," International Journal of Intelligent Systems in Accounting, Finance and Management, special issue on Mathematical and Computational Models of Organizations: Models and Characteristics of Agent Behavior, 1993.
 M. J. Prietula, K.M. Carley and L. Gasser: "A Computational Approach to Organizations and Organizing," in Simulating Organizations, MIT Press, 1998.
T. Sugawara, O. Akashi and S. Kurihara, "Maintenance of Organizational Information in Dynamic Environments," Proceedings of Fourth Pacific Rim International Workshop on Multi-agents (PRIMA2001), Taipei, Taiwan, pp. 293 - 304, 2001.
T. Sugawara, O. Akashi, S. Kurihara and S. Sato, ``An Organization-Related Information Maintenance Component,'' in Proc. of the 4nd Int Conf. on Multi-Agent Systems (ICMAS2000), pp. 443 - 444, 2000.
© Toshiharu Sugawara