An easy-maintenance, reusable approach for building knowledge-based systems: application to landscape assessment

摘要

Multiple Classification Ripple-Down Rules (MCRDR) is an extended methodology which allows an expert to build and maintain a knowledge-based system for multiple classification without technical assistance. Current MCRDR knowledge bases do not exhibit an explicit model of the relationships for the domain terms used by the expert. This is a strong impediment for both reusing and sharing of MCRDR knowledge bases, as well as for rapid development and maintenance. In this work, we describe how a domain knowledge ontological framework can be integrated with MCRDR, so providing this with explicit reusable knowledge components.