行动推理中若干问题的研究

• 作 者: 戈也挺 朱朝晖 陈世福
• 机 构: 南京大学计算机科学与技术系,南京大学计算机科学与技术系,南京大学计算机科学与技术系 南京 210093,南京 210093,南京 210093
• 出 处: 计算机科学
• 发表日期: 2000年03期

1 引言我们所面临的世界是不断动态变化的,一个智能系统往往需要对动态变化的环境做出反应,其中一个重要方面是对各种行动的结果进行预测、推理,以决定下一步的目标和动作。John McCarthy提出进行行动推理(Reasoning about action)研究,并认为行动推理在常识推理中占有基础性的地位。至此以后,行动推理成为人工智能的一项重要研究内容。利用形式化的方法对世界和行动进行描述和推理构成了行动推理的主要内容。行动推理有时也被称为行动逻辑。在这里我们把关于行动和变化的推理总称为行动推理。行动推

1 INTRODUCTION The world we face is constantly evolving. An intelligent system often needs to respond to a dynamically changing environment. One of the important aspects is to predict and reason the outcome of various actions in order to determine the next goal and action. John McCarthy proposed a study of Reasoning about action and argued that action reasoning holds a fundamental position in common-sense reasoning. Since then, action reasoning has become an important study of artificial intelligence. The use of formal methods to describe and reason about the world and actions constitutes the main element of action reasoning. Act inference is sometimes called action logic. Here we generalize the reasoning about action and change as action reasoning. Action push