Database Methods in Expert Systems

近年来，基于规则和基于框架的系统成为开发广泛的知识密集型应用程序的流行工具。尽管这些工具为应用程序开发人员提供了重要的帮助，但是构建专家系统在很大程度上仍然是一项特别的和非常劳动密集型的活动。最近在知识表示和演绎数据库研究方面取得的成就使得尝试基于系统逻辑的专家系统方法的时机成熟。提出的论点是，专家系统编程可以通过将开发人员从指定控制策略的大多数细节的负担中解脱出来，从而使其更具声明性。这个负担应该由系统来承担，系统将使用各种优化技术来实现理想的响应时间。本研究将设计并实现一种集规则与框架于一体，适合于不确定性推理的语言。作为纯粹的声明性语言，这种语言将缩短知识密集型应用程序的开发阶段，并使它们更加可靠和可扩展。工业专家系统开发人员以及知识工程研究人员将受益于本研究的结果。

In recent years rule-based and frame-based systems became popular tools for developing a wide range of knowledge intensive applications. Although these tools provide significant help to the application developer, building expert systems remains a largely ad hoc and a very labor-intensive activity. Recent achievements in knowledge representation and deductive database research made time ripe for attempting a systematic logic-based approach to expert systems. The proposed contention is that expert system programming can be made more declarative by releaving the developer from the burden of specifying most details of control strategy. This burden should be taken over by the system which will use various optimization techniques to achieve desirable response time. In this research, a language which integrates rules and frames under one roof, and is suitable for reasoning with uncertainty will be designed and implemented. Being purely declarative, this language will allow shortening of the development phase of knowledge-intensive applications and make them more reliable and extensible. Industrial expert system developers as well as knowledge engineering researchers will benefit from theresults of this research.