基于模糊贝叶斯网络的软件过时淘汰影响研究

摘要

For the sake of making scientific decision on software obsolescence ,the concept and definition of software obsolescence based on the intrinsic characteristic of software were pointed out ,the software obsolescence mechanism was analyzed and the main influencing attributes of software obsolescence were pointed out ,the main influencing attributes based on the thought of fuzzy Bayesian network and sample distribution were blurred ,the subjective influence factors were eliminated and the condition mutual information and the maximum weight directed tree among all the attribute variables were obtained ,the fuzzy prior proba‐bility and contingent probability were reasoned and the parameter learning method was concluded ,the foundation and reasoning method was validated by the help of the example finally .The result shows that the model is believable and effective .%为对软件的过时淘汰进行科学决策，依据软件的固有特性，提出软件过时淘汰的理念并给出相关定义；分析软件过时淘汰机理，提出软件过时淘汰的主要影响属性，依据模糊贝叶斯网络和抽样分布的思想对主要影响属性进行模糊化处理，消除主观因素的影响并获得所有属性变量之间的条件互信息和最大权重有向树；推理模糊先验概率估计和条件概率估计，归纳参数学习方法。通过案例验证了模型的建立与推理方法，其结果表明该模型具有可信性和有效性。