Continuous process improvement using Bayesian belief networks

Mainstream software quality research has been unable to provide methods which can adequately model or predict software quality. The emergence of Bayesian Belief Networks has made it possible to combine the disparate sources of evidence about process and product needed for a comprehensive judgement about software quality. Bayesian Belief Networks provide an innovative way to increase productivity, quality and hence the competitiveness of a software organisation. Our ultimate goal is to employ Bayesian Belief Networks within a continuous software process improvement setting. In this paper we discuss some of the issues surrounding Bayesian Belief Network software process modelling and outline directions for future research.