Objective software quality assessment

Acceptance of software as being of sufficient quality for use in safety critical systems is a highly subjective process. A modeling technique, Bayesian Belief Networks (BBNs) offers an approach to making these assessments more objective. Use of BBNs can make the assessment process visible and repeatable such that it may be reviewed and validated. The method allows for incorporating data that relates review conclusions to observed quality. Furthermore, it supports combining software engineering measurements with review conclusions in the assessment process. These features offer a path forward to making the software assessment simpler, repeatable, and more objective. This paper introduces BBNs and illustrates the potential application to software assessment