ABSTRACT
Being able to incrementally define and test knowledge bases for intelligent systems is desirable. However, as more knowledge is added, the knowledge engineer must ensure that undesirable interactions between the existing and additional knowledge do not occur. One intelligent monitoring system, the Hazard Monitor (HM), provides for the ability to add knowledge incrementally by including an arbitration process. This paper describes the knowledge requirements and the architecture for Hazard Monitor which facilitates incremental knowledge base development.
