Deriving and Maintaining Rules in an Intelligent DBMS (Computer and Information Science)

本研究探讨了支持智能数据处理的推理规则的自动生成。例如，在基于知识的系统中，产生式规则编码用于专家推理的知识。在数据库系统中，推理规则编码用于语义查询优化的完整性约束。这种基于规则的系统现在依赖于人类专家来提供规则。该方案着眼于数据库应用，开发了一个语义查询优化系统，该系统根据数据库的使用模式和数据库状态的变化，自动派生和维护规则。规则的自动派生是一种很有前途的自适应数据库优化技术，它结合了数据库理论和人工智能的研究领域。这项研究的重要意义在于，它将导致多种基于规则的计算机程序性能的提高。系统派生的规则可能比人类专家开发的规则集更灵活、更高效。此外，为自动评估各种规则集而开发的标准也可能对人类专家有用。

This research investigates the automatic generation of inference rules to support intelligent data processing. For example, in knowledge-based systems, production rules encode knowledge used for expert reasoning. In database systems, inference rules encode integrity constraints used for semantic query optimization. Such rule-based systems now depend on human experts to supply the rules. This proposal focuses on a database application, developing a semantic query optimization system that automatically derives and maintains rules, based on database usage patterns and changes to the database state. The automatic derivation of rules is a promising self-adaptive database optimization technique that combines areas of research from both database theory and artificial intelligence. The importance of this research is that it will lead to improvements in the performance of many kinds of rule-based computer programs. System-derived rules can be more flexible and efficient than a set developed by a human expert. In addition, the criteria developed for automatic evaluation of various rule sets may be of use to human experts as well.