结合经验数据和专家知识,基于贝叶斯网建立了软件体系结构层次的结构特征、变化原因与软件适应性之间的因果关联模型,即软件体系结构层次的软件适应性预测模型,扩展贝叶斯网学习算法解决了该预测模型中较弱因果关系的发现问题.最后给出应用预测模型在软件体系结构层次上评估软件适应性的方法和实例.

In this paper, the prediction model based on Bayesian is constructed for presenting cause and effect relationships between software structure features and software flexibility by consulting the experiential data and expert knowledge. In addition, the Bayesian networks learning technique is extended to find out the weak causal relationships in the software flexibility prediction model constructing. Finally the method and examples using the prediction model to assess software flexibility in the software architecture level are presented.

Supported by the National Natural Science Foundation of China under Grant Nos.60773155, 90818017 (国家自然科学基金)