A hypothetical application of the Statistical Modeling and Estimation of Reliability Functions for Software (SMERFS) to Knowledge-Based Systems (KBSs)

摘要

Software reliability models are applied to conventional software to compute several important quantities. Two of these are the rate at which software fails and the number of software faults remaining in software. The software failure rate and the number of faults remaining in software are measures of software trustworthiness or reliability. Software reliability models, however, have never been applied to knowledge-based systems to determine these quantities. This paper reports a hypothetical application of the Statistical Modeling and Estimation of Reliability Functions for Software (SMERFS) to knowledge-based systems. SMERFS, developed by the Naval Surface Warfare Center (NSWC), contains a representative set of software reliability models “chosen for their performance in comparative studies and their ability to handle data collected from various testing environments.”