Applying bayesian belief networks for early software quality modeling

Bayesian Belief Networks (BBNs) are becoming popular within the Software Engineering research community. This report analyses the applicability of Bayesian Belief Nets to model quality of evolutionary software systems. The analysis is done by giving an overview of the domains where BBNs are currently applied. Then a process is introduces to apply BBNs in software quality modeling. After that a discussion on the applicability is conducted. We conclude that Bayesian Belief Networks are profitable for modeling software quality of evolutionary software systems, because the complexity of such models can be made transparent with BBNs. However, BBNs could not be practical to apply for large models because of the great effort to build initial network. This report is written as a contribution to Work package 3.5.2 of the ITEA EMPRESS project, in which Fraunhofer IESE was participating