摘要
对软件可靠性早期预测模型的研究,首先必须考虑影响软件可靠性的主要因素。而影响因素权重的确定可以说是整个早期预测模型的"灵魂"。针对主观赋权法和客观赋权法的优缺点,提出了基于神经网络和MIV的组合赋权法来分析软件可靠性的影响因素。该方法首先根据主观赋权法得到影响因素的排序,然后利用神经网络的强大的非线性处理能力以及MIV进行客观分析,从而获得更客观更具解释性的权重值和排序。
The main factors affecting software reliability must be considered firstly before early software reliability prediction model is studied.Determination of the factors weights can be said the ＂soul＂ of early prediction model.To the question about the advantages and disadvantages of the objective weighting method and the subjective weighting method, a combination weighting method based on BP neural network and MIV is proposed to analyze factors affecting software reliability.In this method,the sorting factors are got by subjective method firstly,and then the neural network with big and powerful non- linear processing ability and MIV is used as objective analysis tools.At last the more objective and more explanatory weight value and sorting are got.
出处
《计算机工程与应用》
CSCD
北大核心
2010年第33期58-62,共5页
Computer Engineering and Applications
基金
四川省科技计划项目(No.2008GZ0003)
关键词
可靠性因素
BP网络
平均影响值
权重分析
factors affecting software reliability
BP network
Mean Impact Value（MIV）
weight analysis
作者简介
E—mail：kylunwen@126．com 吴洪丽（1976-），女，博士研究生，讲师，CCF会员，主要研究领域：软件工程理论、数据挖掘；
蒋永辉，通讯作者，男，硕士研究生，主要研究领域：软件过程技术与方法、文本挖掘。
