Strategic reuse framework for high-performance knowledge bases development

Abstract

In recent years, researchers from knowledge-based system (KBS) fields have pushed toward rapidly develop a reusable knowledge base with a greater flexibility and adaptability to empower domain experts developing their own domain-oriented high performance knowledge-based systems. Such systems should have the strategic capabilities of dynamically adapting their performance parameters, improve their knowledge base or method of operation in order to accomplish a set of tasks. Our research intends to develop a strategic-reuse system that takes as input a declarative description of a domain, and then automatically generates as output an efficient, domain-specific KBS development environment along with a reusable knowledge base library. A test plan can be further developed based on the relationships between the specific KBS performance requirement (such as time, memory, or information quality) and the KB components' strategic characteristics. This paper focuses on developing such a strategic-reuse framework to characterize KB components in terms of four generic strategic features: Computational Efficiency, Interactive Capability, Modeling Flexibility, and Knowledge Usability. An analogy of structural-behavioral-functional-teleological (SBFT) multimodeling framework for diagnosis problem solving is adopted as a cognitive base to rationalize the complex strategic interconnections among those knowledge base components.