organized into a form from which the user can effectively and efficiently extract required information. The service application software agents collaborate with other software agents to achieve general goals set by users, and based on user profiling, generate pertinent situation changes that may be of interest to the user. The agents support automatic, dynamic, adaptive allocation of transport and processing resources, and replicate as necessary for efficiency and to ensure continuity of services provided to the user.

Intelligent application software agents must provide an array of functions appropriate to the user's mission and situation, and exchange information and status with other application software agents to provide integrated yet distributed execution of requested user services. These agents automatically select and perform their functions depending on specific user requirements and profiled user interest areas. The agents provide discovery and integration of text, tabular and geospatial data from multiple, heterogeneous databases, broker between other agents for sharing of information, and negotiate with service agents to establish appropriate network and resource allocations to achieve their goals. These agents are adaptive, in that they profile user needs for information such as measurements, targets, maps, changes in areas, and models against direct user input, past user requirements, and an understanding of user mission, status, and intentions.

Although successful examples of both types of agent exist, there is general agreement that more investment to strengthen the technology base is needed before robust agents can be routinely constructed. This is not trivial. As the nation attempts to integrate DOD and commercial geospatial data, many important questions remain open. Needed technologies include:

  • Universal language and computational models for declaring agents,

  • Representation technology for knowledge and system resources,

  • Algorithms and protocols for agent management and interagent negotiation and information exchange, and

  • Automated learning and user-profiling techniques.

Because of commercial interest, DOD need not pay the entire bill, but the pattern in the past has been one of DOD investment in high-risk developments and commercial investment in turning the successful developments into products. Even if this pattern is broken, some DOD investment is needed in domain-specific developments.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement