A modularization approach for PROLOG knowledge bases

摘要

One of the most critical disadvantages of knowledge based systems is their slow execution speed. Knowledge based systems run about ten to one hundred times slower than traditional (procedural) information systems. This has generally limited their application environments and constrained the development of hybrid systems requiring both traditional and knowledge based approaches. This paper presents a framework for modularizing PROLOG knowledge bases such that the inferencing process to execute a goal is localized to an independent module consisting of only a subset of the knowledge base. Note that the motivation for modularization in a knowledge based system is quite different from the motivations for modularization in a traditional system. The presented modularization approach is expected to contribute to the development of more efficient knowledge based systems and more effective hybrid systems requiring both knowledge based and traditional approaches.