AIID/Hats: An Architecture for the Interpretation of Intelligence Data

AIID/Hats：解释情报数据的架构paul Cohen，南加州大学信息科学研究所意外事件研究中心。该提案将展示贝叶斯黑板的新技术如何扩展到涉及数千个人的模糊和不完整数据的问题。意外事件研究中心（CRUE）开发了一个贝叶斯黑板系统的原型，名为AIID，即智能数据解释架构。贝叶斯黑板结合了黑板系统和贝叶斯信念网络的技术。这种结合的一个结果是，对世界上的情报收集和黑板上的推理的控制可以是理性的，也就是说，基于概率论和效用论。这个项目的第二部分是一个名为Hats的恐怖活动模拟器。模拟器是数千名代理人（“帽子”）的家，他们前往会议。有些帽子是隐蔽的恐怖分子，少数帽子是已知的恐怖分子。所有的帽子都由计划者生成的计划来管理。恐怖分子的计划以破坏地标而告终。与Hats模拟器对抗的游戏目标是在恐怖分子执行计划之前找到他们。一种是支付有关帽子的信息，一种是支付虚假逮捕和毁坏地标的信息。在游戏结束时，玩家会得到一个分数，也就是这些成本的总和。我们的目标是理性地玩Hats，即以最少的信息、虚假逮捕和破坏地标的综合成本抓住恐怖组织。贝叶斯黑板是最好的基于知识的信息融合体系结构和用概率假设进行组织和推理的最好方法的结合。贝叶斯黑板，因为它给假设提供了一个合理的概率语义，使理性智力分析理论成为可能，或者，同样地，发展了智力效用的数学公式和针对嫌疑人的行动。该项目是USC信息科学研究所的意外事件研究中心（CRUE）开发理性智能分析工具的一部分。CRUE的科学重点是数学和模拟模型，这些模型统一了入侵检测和生物监视等明显不同的活动。这些模型合在一起构成了一个叫做城市安全信息学的新领域。这些模型将在AIID/Hats系统中进行测试，AIID将被开发为情报分析人员的工具。目前的提议和CRUE的所有努力都是为情报界服务的；在本例中是通过KD-D程序。

AIID/Hats: An Architecture for Interpreting Intelligence DataPaul Cohen, Center for Research on Unexpected Events, USC Information Sciences Institute.This proposal will demonstrate how the new technology of Bayesian blackboards scales to problems involving ambiguous and incomplete data about thousands of individuals. The Center for Research on Unexpected Events (CRUE) has developed a prototype of a Bayesian blackboard system called AIID, Architecture for the Interpretation of Intelligence Data. A Bayesian blackboard combines the technologies of blackboard systems and Bayesian belief networks. One consequence of this marriage is that the control of intelligence gathering in the world and inference on the blackboard can be rational, that is, grounded in probability and utility theory.The second part of this project is a simulator of terrorist activities called Hats. The simulator is home to thousands of agents ("hats"), which travel to meetings. Some hats are covert terrorists and a very few hats are known terrorists. All hats are governed by plans generated by a planner. Terrorist plans end in the destruction of landmarks. The object of a game against the Hats simulator is to find terrorist task forces before they carry out their plans. One pays for information about hats, and also for false arrests and destroyed landmarks. At the end of a game, one is given a score, which is the sum of these costs. The goal is to play Hats rationally, that is, to catch terrorist groups with the least combined cost of information, false arrests, and destroyed landmarks.The Bayesian blackboard is a marriage of the best knowledge-based information fusion architecture and the best methods to organize and reason with probabilistic hypotheses.The Bayesian blackboard, because it gives a sound probabilistic semantics to hypotheses, enables a theory of rational intelligence analysis, or, equivalently, the development of mathematical formulations of the utility of intelligence and action against suspects.This project is part of an effort by the Center for Research on Unexpected Events (CRUE) at USC's Information Sciences Institute to develop tools for rational intelligence analysis. CRUE's scientific focus is mathematical and simulation models which unify apparently diverse activities such as intrusion detection and biosurveillance. Taken together these models constitute a new field called informatics for urban security. These models will be tested in the AIID/Hats system, and AIID will be developed as a tool for intelligence analysts. The current proposal, and all CRUE efforts, serve the intelligence community; in this instance through the KD-D program.