Control Issues in Asynchronous Parallel Knowledge-Based AI Programs

9321324 较小的控制，即智能选择要执行的操作，对于基于知识的系统（例如计划人员、作业车间调度程序和语音理解系统）的性能至关重要。 对于并行系统尤其如此，其中控制代理具有最大化并行资源的利用的额外目标。 所提出的研究通过基于任务交互和环境影响分析的控制理论的发展来研究并行基于知识的系统中的控制问题。 控制问题的理论方法基于以下观察：控制是一种基于知识的知识，基于对问题解决过程、问题域和问题解决事件发生的特定环境之间相互作用的理解。 这项工作将是先前工作的延伸，旨在开发一个领域结构如何影响的正式模型，并限制用于在该领域内开发解决方案和开发环境随机模型的方法的选择。 这些努力将扩展到在并行计算中特别重要的建模效果。 具体来说，将探讨任务交互的本质。

9321324 Lesser Control, the intelligent selection of actions to execute, is critical to the performance of knowledge-based systems such a planners job shop schedulers, and speech understanding systems. This is particularly true for parallel systems in which the controlling agent has the additional goals of maximizing the utilization of parallel resources. The proposed research studies the issues of control in parallel knowledge-based systems through the development of a theory of control based o an analysis of task interactions and the influence of the environment. The theoretical approach to control issues is based on the observation that control is a knowledge-based on an understanding of the interactions between the problem-solving process, the problem domain, and the specific environment in which a problem-solving episode takes place. This effort will be an extension of previous efforts to develop a formal model of how the structure of a domain influences ad constrains the choice of methods used to develop solutions within that domain and in developing stochastic models of environments. These efforts will be extended towards modeling effects that are particularly significant in parallel computations. Specifically, the nature of task interactions will be explored.