Research And Implementation Of Software Risk Management Model Based On Bayesian Networks

Software development project is one of the most important engineering projects in our society, but a mass of risks existing in the procedure of project development give the project rise to failure. The technology of software development improves constantly, and there are a lot of methods and models of risk management, but still can't control the risk effectively. Directing to this fact, introducing an effective mathematical reasoning model, combing with risk management, introducing a software project risk management model based on Bayesian network.Directing to feature of the procedure of risk management, analyzes some classical risk management models and the structure of Bayesian network, proposing a software project risk management model based on Bayesian network. The model consists of the following items: model of risk Database, model of identifying Bayesian network risk, model of analyzing Bayesian network risk, model of assessing risk, model of handling risk and model of monitoring risk. Firstly, identify risks exist in project according to historical risk library, and the list of risk is composed of these risks; construct the structure of Bayesian network based on the former work, perfect the structure of Bayesian network according to the project condition; analyze the project data of every stage of development continuously and predict risk using the reasoning algorithm of Bayesian network; lastly, calculate risk equivalent, provide risk assessment, adopt appropriate measures to reply risk according to risk assessment. In proposed model, presenting the procedure of software project risk management based on Bayesian network and the major management activities, and introducing model of risk database which can use historical risk library, user can study case conveniently. Proposing executing risk management according to the procedure of software development, participants of software project can focus risk on every development stage. Design implementation scheme according to the model and implement the model. The application of our system can predict the probability of risk effectively, can analyze the relationship between risks and the influence of risks, so participants of software project can adopt appropriate measures and adjust work under the direction of the risk assessment, so the project can accomplish on plan. Our system provide a new technical approach and implementation for predicting, analyzing and controlling software project risk, and our research has a relative high research meaning and application value.