Bayesian Network-based Software Risk Management Issues

The risk management is a very essential part in Software project management. However, due to the lack of enough understanding, many organizations neglect the software project risk management.Firstly, this article introduces the typical theory about risk management and the typical risk management model, which is simplified for the real problem later.The main problems in software risk management are pointed out. The limitations of existing management methods to handle these problems are analyzed, an method using Bayesian networks to conduct risk prediction is presented, and the local and synthetic models are individually given. The algorithms are given to deal with situations when the new evidences are captured.An application example shows that this method is effective to predict the risk probabilities during the specification and design phases of the project.