Intelligent Information Gathering Using Decision Models

调查协调信息收集。今天，国家信息基础设施提供了大量信息，有很大的潜力来提高决策质量和信息消费者的生产力。预计在今后5-10年内，可获得的信息量和来源数量将迅速增加。将有改进的信息检索和信息提取技术。目前出现的大量信息源及其不同级别的可访问性、可靠性和相关成本提出了一个复杂的信息收集协调问题，如果没有高级别的信息过滤，人类决策者不可能解决这个问题。为了能够利用NII的巨大潜力，我们建议开发信息收集任务自动化的下一个逻辑步骤：根据其对用户决策或任务质量的边际贡献来组织信息收集活动。这一方法强调，信息收集是决策过程的中间步骤。本研究将提供一个系统，用户的决策模型的一个明确的表示，使信息收集活动可以组织的基础上，其对决策质量的影响。自动化的信息收集规划任务，研究将调查一个创新的扩展标准的信息价值理论，解决信息的可靠性，检索成本，实时操作和不确定性的问题。当在资源约束（与通信和数据库访问的成本、有限的计算能力和有限的时间量有关）下操作时，这种方法应该比当前技术实现显著的性能改进

To investigate coordinated information gathering. The vast amount of information available today on the National Information Infrastructure has a great potential to improve the quality of decisions and the productivity of consumers of this information. Rapid expansion of the amount of available information and the number of sources is expected over the next 5-10 years. Improved information retrieval and information extraction technologies will become available. The large number of information sources that are currently emerging and their different levels of accessibility, reliability, and associated costs present a complex information gathering coordination problem that a human decisionmaker cannot possibly solve without high-level filtering of information. To be able to exploit the great potential of the NII we propose to develop the next logical step in automation of information gathering tasks: to organize information gathering activity based on its marginal contribution to the quality of the user's decision or task. This approach emphasizes the fact that information gathering is an intermediate step of a decisionmaking process. This research will provide a system with an explicit representation of the user's decision model, so that information gathering activity can be organized on the basis of its effect on the quality of the decision. To automate the information gathering planning task, the research will investigate an innovative extension to the standard value-of-information theory, that addresses the problems of information reliability, retrieval costs, real-time operation, and uncertainty. When operating under resource constraints (related to the cost of communication and database access, limited computation power, and limited amount of time) this approach should achieve a significant performance improvement over current techniques